diff --git a/docs/benchmarks/euroc-baseline.mdx b/docs/benchmarks/euroc-baseline.mdx
index 582949c..3905a14 100644
--- a/docs/benchmarks/euroc-baseline.mdx
+++ b/docs/benchmarks/euroc-baseline.mdx
@@ -55,14 +55,14 @@ The numbers come from `scripts/euroc-baseline.json`, which
| MH_01_easy | 0.331 | 0.333 | 0.096 | 0.072 | 3.5× |
| MH_02_easy | 0.456 | 0.393 | 0.168 | 0.117 | 2.3× |
| MH_03_medium | 0.500 | 0.472 | 0.147 | 0.096 | 3.2× |
-| MH_04_difficult | 0.375 | 0.379 | 0.147 | 0.127 | 2.6× |
-| MH_05_difficult | 0.671 | 0.629 | 0.458 | 0.281 | 1.4× |
-| V1_01_easy | 0.200 | 0.197 | 0.069 | 0.063 | 2.8× |
-| V1_02_medium | 0.163 | 0.162 | 0.058 | 0.048 | 2.8× |
-| V1_03_difficult | 0.158 | 0.161 | 0.067 | 0.059 | 2.4× |
-| V2_01_easy | 0.223 | 0.215 | 0.127 | 0.060 | 1.7× |
-| V2_02_medium | 0.338 | 0.334 | 0.073 | 0.044 | 4.6× |
-| V2_03_difficult | 0.173 | 0.175 | 0.181 | 0.087 | 1.0× |
+| MH_04_difficult | 0.384 | 0.385 | 0.147 | 0.127 | 2.6× |
+| MH_05_difficult | 0.686 | 0.640 | 0.458 | 0.281 | 1.4× |
+| V1_01_easy | 0.192 | 0.188 | 0.069 | 0.063 | 2.7× |
+| V1_02_medium | 0.155 | 0.153 | 0.058 | 0.048 | 2.6× |
+| V1_03_difficult | 0.145 | 0.147 | 0.067 | 0.059 | 2.2× |
+| V2_01_easy | 0.191 | 0.184 | 0.127 | 0.060 | 1.5× |
+| V2_02_medium | 0.327 | 0.322 | 0.073 | 0.044 | 4.4× |
+| V2_03_difficult | 0.174 | 0.177 | 0.181 | 0.087 | 1.0× |
The ratio compares cortex's window with OpenVINS mono; both are monocular.
@@ -75,14 +75,14 @@ Over the window:
| MH_01_easy | 0.370 m / 1.57° | 0.137 m / 1.55° | 0.141 m / 1.67° |
| MH_02_easy | 0.478 m / 1.35° | 0.173 m / 1.07° | 0.132 m / 0.86° |
| MH_03_medium | 0.386 m / 0.75° | 0.178 m / 0.61° | 0.159 m / 0.62° |
-| MH_04_difficult | 0.548 m / 1.03° | 0.206 m / 0.83° | 0.213 m / 0.50° |
-| MH_05_difficult | 0.701 m / 0.95° | 0.506 m / 1.06° | 0.300 m / 0.85° |
-| V1_01_easy | 0.330 m / 4.04° | 0.262 m / 2.84° | 0.229 m / 2.83° |
-| V1_02_medium | 0.209 m / 2.31° | 0.110 m / 0.57° | 0.106 m / 0.73° |
-| V1_03_difficult | 0.235 m / 2.56° | 0.125 m / 1.24° | 0.125 m / 1.02° |
-| V2_01_easy | 0.447 m / 2.41° | 0.220 m / 1.06° | 0.102 m / 0.89° |
-| V2_02_medium | 0.301 m / 4.09° | 0.090 m / 1.55° | 0.061 m / 1.27° |
-| V2_03_difficult | 0.225 m / 1.74° | 0.201 m / 1.56° | 0.108 m / 1.38° |
+| MH_04_difficult | 0.548 m / 0.76° | 0.206 m / 0.83° | 0.213 m / 0.50° |
+| MH_05_difficult | 0.722 m / 0.87° | 0.506 m / 1.06° | 0.300 m / 0.85° |
+| V1_01_easy | 0.318 m / 4.02° | 0.262 m / 2.84° | 0.229 m / 2.83° |
+| V1_02_medium | 0.208 m / 1.87° | 0.110 m / 0.57° | 0.106 m / 0.73° |
+| V1_03_difficult | 0.207 m / 1.93° | 0.125 m / 1.24° | 0.125 m / 1.02° |
+| V2_01_easy | 0.370 m / 1.94° | 0.220 m / 1.06° | 0.102 m / 0.89° |
+| V2_02_medium | 0.272 m / 3.03° | 0.090 m / 1.55° | 0.061 m / 1.27° |
+| V2_03_difficult | 0.222 m / 1.50° | 0.201 m / 1.56° | 0.108 m / 1.38° |
## What the baseline says
@@ -256,6 +256,57 @@ turns that into a pass/fail check that any backend can be run through:
collapse.
- Every verdict below is the same at any bound between 0.13 and 0.9.
+### Results with the tilt–bias seed prior (#609)
+
+Recorded at 7a38162 for both backends, median over the three seeds. #609 made a
+static seed's P₀ correlate tilt with accelerometer bias as the window implies
+(σ_tilt = σ_ba / g, σ_ba = 0.2 m/s²), and seed the bias along up from
+|f| − g. Gravity-aligned and dynamic seeds keep the isotropic prior
+([S1](../pipeline/stages/s1-initialization.mdx)).
+
+| Gate condition | MSCKF | R-IEKF |
+|---|---|---|
+| 1. No divergence (66 matched runs) | **fail:** the same 2 (MH_03 1 px seed 2, 26.8 m; MH_04 1 px seed 3, 8.8 km) | **fail:** the same 2 (2.3 m; 9.0 km) |
+| 2. Matched beats default (median, 1 px / 0.5 px) | **pass:** 0.36× vs 0.64×; 0.25× vs 0.58× (0.45/0.52, 0.33/0.53 before) | **pass:** 0.32× vs 0.53×; 0.19× vs 0.40× (0.40/0.41, 0.18/0.36 before) |
+| 3. No yaw collapse (median yaw σ min ÷ anchor ≥ 0.5) | **fail:** 0.152 (0.179 before) | **pass:** 0.969 (0.984 before) |
+| Median NIS, matched runs | 1.00 | 1.00 |
+| Roll/pitch NEES, matched runs that don't diverge | 1.0–4.8 (1.1–24 before) | 0.9–12.6 (0.9–6.8 before) |
+
+At 0.5 px, matched `R`, per sequence (window ATE ratio to OpenVINS mono, yaw σ
+min ÷ anchor, NIS):
+
+| Sequence | MSCKF ratio | MSCKF yaw σ | MSCKF NIS | R-IEKF ratio | R-IEKF yaw σ | R-IEKF NIS |
+|---|---:|---:|---:|---:|---:|---:|
+| MH_01_easy | 0.4× | 0.09 | 1.00 | 0.4× | 0.84 | 1.00 |
+| MH_02_easy | 0.4× | 0.17 | 1.00 | 0.2× | 0.90 | 1.00 |
+| MH_03_medium | 0.3× | 0.17 | 1.00 | 0.2× | 0.89 | 1.00 |
+| MH_04_difficult | 0.2× | 0.16 | 1.00 | 0.2× | 0.91 | 1.00 |
+| MH_05_difficult | 0.1× | 0.19 | 1.00 | 0.0× | 0.91 | 1.00 |
+| V1_01_easy | 0.3× | 0.16 | 1.00 | 0.3× | 0.81 | 1.00 |
+| V1_02_medium | 0.4× | 0.21 | 1.00 | 0.2× | 0.90 | 1.00 |
+| V1_03_difficult | 0.2× | 0.16 | 1.00 | 0.2× | 0.90 | 1.00 |
+| V2_01_easy | 0.2× | 0.14 | 1.00 | 0.2× | 0.64 | 1.00 |
+| V2_02_medium | 0.2× | 0.15 | 1.00 | 0.2× | 0.67 | 1.00 |
+| V2_03_difficult | 0.1× | 0.15 | 1.00 | 0.1× | 0.82 | 1.00 |
+
+- **The MSCKF's tilt is consistent now.** Its worst roll/pitch NEES among the
+ runs that don't diverge fell from 24 to 4.8, and matched `R` beats the
+ default by a wider margin at both noise levels.
+- **The R-IEKF's yaw σ falls further on the V2 sequences** (0.95 → 0.64 on
+ V2_01 and 0.94 → 0.67 on V2_02 at 0.5 px), and its worst roll/pitch NEES
+ rose from 6.8 to 12.6. Its median still passes condition 3. The suspicion is
+ that the bias couples a correlated prior into yaw through propagation, which
+ the invariant error doesn't make exact (#620).
+- **Neither divergence was the tilt seed.** On MH_04 (1 px, seed 3) the dynamic
+ seed was 0.14° from the truth's tilt; its speed was 0.05 m/s against the
+ truth's 0.42, and the filters explain the missing motion by tilt (3° after
+ 1 s, 117° at the end). MH_03 (1 px, seed 2) is gravity-aligned and seeds zero
+ velocity on a moving start. Both are velocity seeds of moving starts.
+- **A looser prior on the gravity-aligned seed made it worse.** Its window
+ accelerates (|f| − g = 2.6 m/s² on MH_03), so the honest tilt σ is 15–16°.
+ With it three of 66 diverged per backend, MH_03 0.5 px seed 1 at 3.9–5.2 km,
+ and the gravity-aligned seed kept the isotropic prior.
+
### Results with one truth, and S2 integrating both samples (#616)
Recorded at f1551cc for both backends, median over the three seeds. #616 made S2
@@ -304,7 +355,8 @@ min ÷ anchor, NIS):
R-IEKF runs the MSCKF's initializer since #608), and both reach a roll/pitch
NEES of 34,000–40,000 and about 9 km: the seed, not either filter. MH_03
(1 px, seed 2) starts by gravity alignment; the MSCKF reaches 26.8 m, the
- R-IEKF 2.3 m. Both point at the tilt and bias seed of #609.
+ R-IEKF 2.3 m. #609 first suspected the tilt and bias seed; measured, neither
+ is the tilt seed: both are moving starts' velocity seeds (see above).
### Results with one truth, before #616
@@ -521,8 +573,48 @@ Per sequence, window 4-DoF ATE (m) and window rotation RPE (°):
- **Its NIS is now reported.** It averages 0.97–2.55 on the Machine Hall and on
V1_01 and V2_01, and 6.8–15.4 on V1_02, V1_03, V2_02 and V2_03, where the filter
is over-confident or the gate admits outliers.
-- **Roll/pitch is still wrong at the start:** V1_01 has a roll/pitch NEES near
- 3000 in every configuration, the tilt and accelerometer-bias seed of #609.
+- **Roll/pitch is still wrong at the start (before #609):** V1_01 has a
+ roll/pitch NEES near 3000 in every configuration. #609 changed the seed's
+ prior; the 2.4° gap to V1_01's ground truth stays, for the reason below.
+
+### After #609: the tilt–bias seed prior
+
+On the real KLT front end, window 4-DoF ATE (m) and window rotation RPE (°),
+#616 → #609:
+
+| Sequence | MSCKF ATE | R-IEKF ATE | OpenVINS mono | MSCKF ÷ mono | R-IEKF ÷ mono | MSCKF rot | R-IEKF rot |
+|---|---:|---:|---:|---:|---:|---:|---:|
+| MH_01_easy | 0.333 → 0.333 | 0.312 → 0.312 | 0.096 | 3.5× | 3.2× | 1.57 → 1.57 | 0.82 → 0.82 |
+| MH_02_easy | 0.393 → 0.393 | 0.370 → 0.370 | 0.168 | 2.3× | 2.2× | 1.35 → 1.35 | 1.37 → 1.37 |
+| MH_03_medium | 0.472 → 0.472 | 0.402 → 0.402 | 0.147 | 3.2× | 2.7× | 0.75 → 0.75 | 0.66 → 0.66 |
+| MH_04_difficult | 0.379 → 0.385 | 0.359 → 0.360 | 0.147 | 2.6× | 2.4× | 1.03 → 0.76 | 0.46 → 0.47 |
+| MH_05_difficult | 0.629 → 0.640 | 0.612 → 0.614 | 0.458 | 1.4× | 1.3× | 0.95 → 0.87 | 0.68 → 0.68 |
+| V1_01_easy | 0.197 → 0.188 | 0.182 → 0.181 | 0.069 | 2.7× | 2.6× | 4.04 → 4.02 | 3.74 → 3.69 |
+| V1_02_medium | 0.162 → 0.153 | 0.146 → 0.144 | 0.058 | 2.6× | 2.5× | 2.31 → 1.87 | 0.81 → 0.82 |
+| V1_03_difficult | 0.161 → 0.147 | 0.131 → 0.131 | 0.067 | 2.2× | 2.0× | 2.56 → 1.93 | 0.88 → 1.11 |
+| V2_01_easy | 0.215 → 0.184 | 0.219 → 0.217 | 0.127 | 1.4× | 1.7× | 2.41 → 1.94 | 1.33 → 1.62 |
+| V2_02_medium | 0.334 → 0.322 | 0.261 → 0.259 | 0.073 | 4.4× | 3.5× | 4.09 → 3.03 | 1.37 → 1.48 |
+| V2_03_difficult | 0.175 → 0.177 | 0.177 → 0.177 | 0.181 | 1.0× | 1.0× | 1.74 → 1.50 | 1.80 → 1.89 |
+| **median** | | | | **2.58× → 2.62×** | **2.44× → 2.45×** | **1.74 → 1.57** | **0.88 → 1.11** |
+
+MH_01, MH_02 (dynamic) and MH_03 (gravity-aligned) start from seeds the change
+doesn't touch, and don't move.
+
+- **The MSCKF's rotation drift falls 19–26% in the Vicon rooms** (V2_02 4.09° →
+ 3.03°, V1_03 2.56° → 1.93°) and its V2_01 ATE 14%. Nothing is worse beyond the
+ 10% tolerance; the recorded baseline (`scripts/euroc-baseline.json`) is now
+ 7a38162.
+- **The R-IEKF's rotation drift rises on V1_03 (+26%), V2_01 (+22%), V2_02
+ (+8%) and V2_03 (+5%)**, its median 26% (0.88° → 1.11°), its ATE flat. That
+ is the same family as its gate's
+ yaw σ on V2 (#620).
+- **V1_01 is still 2.4° from its ground truth's tilt**, and the tilt probe
+ (`vio_pipeline --out` writes `tilt.csv`) shows why it can't be closed by the
+ seed: both filters settle at b_a y ≈ 0.13 m/s² with σ_ba = 0.2, with 1.0 and
+ with the isotropic prior, the value they reach on every sequence and the
+ ground truth gives on all but V1_01 (0.55). The gap is a fixed 2.5° rotation
+ in the body frame. The ground truth's bias is its own estimator's output,
+ so the data supports the filter.
### After #616: S2 integrates both samples
diff --git a/docs/first-principles/overview.mdx b/docs/first-principles/overview.mdx
index 88bb476..282b512 100644
--- a/docs/first-principles/overview.mdx
+++ b/docs/first-principles/overview.mdx
@@ -19,7 +19,7 @@ here goes wrong in any MSCKF-style pipeline, and a unit-test suite that passes d
measurement that decided its cause, and how it was fixed.
- [A symptom → cause table and twenty-two pitfalls](signatures.mdx), so the next person
+ [A symptom → cause table and twenty-five pitfalls](signatures.mdx), so the next person
recognizes a failure on sight.
diff --git a/docs/first-principles/regression-suite.mdx b/docs/first-principles/regression-suite.mdx
index 2aad5ba..73300a8 100644
--- a/docs/first-principles/regression-suite.mdx
+++ b/docs/first-principles/regression-suite.mdx
@@ -47,6 +47,7 @@ cortex test case that implements it (Catch2 case names, under `tests/`).
| B3 | S1: SfM under realistic noise | The window builds at 1–2 px and rejects about 0.05 px of pseudo-parallax | "S1 SfM sweep: … 2 px of noise never builds" | characterization |
| B4 | S1: known-answer scale with a lever arm | Exact data recovers the scale to rounding (now 0.78% off) | init-window known-answer fixture, 2% tolerance | weak |
| B5 | S1: bootstrap pair selection | The window builds at 20 Hz where adjacent pairs fail | "S1 SfM bench: the alternatives on the same windows" | locked |
+| B5a | S1: the seed's prior says what its window determines | A static seed's tilt and accelerometer bias are correlated as the window implies (P_θb = σ_ba²/g·[û]ₓ); over 600 seeds their 5-dof ANEES is 5 ± 0.5; a 0.5 m/s² bias is learned once the motion rotates | `tests/sdk/tilt_bias_prior.cpp`; the adapter (`tools/include/branes/tools/invariant_backend_adapter.hpp`) carries the block into the R-IEKF (`tests/tools/invariant_backend_adapter.cpp`) | locked |
| B6 | S2: propagation preserves the gauge | Φ·N = N each step, in both error states; the invariant Φ doesn't depend on the estimate; the body-frame Φ is the exact Jacobian of the mean it propagates (finite differences, 1e-8) | "R-IEKF Phase A: the invariant propagation Phi is state-independent"; "S2 bench sweep: the mean's own Jacobian leaks no yaw, a truncated Phi does"; "S2 interval: Phi is the finite-difference Jacobian of the RK4 mean" | locked (body-frame since ) |
| B6a | S2: the mean closes on a continuous truth | Integrating a fast spline's own noise-free IMU (1.3 rad/s roll and pitch) over a 0.55 s clone window stays within what the camera resolves (a tenth of 0.5 px: 1.09e-4 rad, 0.22 mm at 2 m) and converges at second order; holding one sample over the interval misses it 28× at first order | `gt_spline.cpp`: "S2 closure, the shipped S2 within what the camera resolves (#616)" and its companions | locked |
| B7 | S2: process noise | Predicted covariance matches a Monte-Carlo within 1% per block, 20 m from the origin; cross-correlations within 0.06 | "… predicted covariance matches a Monte-Carlo far from the origin" | locked |
@@ -77,7 +78,7 @@ cortex test case that implements it (Catch2 case names, under `tests/`).
|---|---|---|---|---|
| D1 | Monte-Carlo consistency | 21 seeds or more. Every seed: RMS below 0.15 m, normalized full NEES in (0.25, 3), roll/pitch below 6, yaw below 3, position below 4. Across the seeds: median RMS below 0.06 m | CI: "tracks and stays consistent on every seed" (4 seeds). Manual: the hidden "21-seed sweep" (`[.sweep]`) | partial: 4 seeds locked (R-IEKF), 21 manual |
| D2 | Slow-start robustness | The D1 bands at onsets of 1 s and 2 s | CI: "a slow 2 s start stays consistent" (2 seeds at 2 s). Manual: the hidden "21-seed sweep at slow onsets" (1 s and 2 s) | partial: 2 s on 2 seeds locked (R-IEKF); 1 s and 21 seeds manual |
-| D3 | A consistency control under noise, never noise-free | Pose NEES about dof under 1 px plus the nominal IMU noise, **for every backend** | R-IEKF: D1. Body-frame: "S10 matched-noise control: the body-frame MSCKF leaks information into yaw" pins the leak (median pose NEES above 9, yaw σ falling) | locked (R-IEKF); characterization (body-frame) |
+| D3 | A consistency control under noise, never noise-free | Pose NEES about dof under 1 px plus the nominal IMU noise, **for every backend** | R-IEKF: D1. Body-frame: "S10 matched-noise control: the body-frame MSCKF leaks information into yaw" pins the leak (median yaw NEES above 1.5, yaw σ falling; the median pose NEES until made roll/pitch consistent) | locked (R-IEKF); characterization (body-frame) |
| D4 | Every arithmetic type end to end | float, posit32, long double: P PSD every frame; positions within 1000·max(ε_type, ε_double) metres of the double result | "the estimator runs end to end in every arithmetic type" | locked |
| D5 | Genericity audit | A strict type with explicit-only conversions matches double bit for bit | `strict_double.cpp` | locked |
| D6 | Per-stage precision | One stage in float within 1000 float ulps; PSD checked at the narrowest stage's ε | `stage_types.cpp` | locked |
@@ -94,7 +95,7 @@ cortex test case that implements it (Catch2 case names, under `tests/`).
| E6 | Mixed-precision presets on real data | No divergence for any shipped stage-type preset; ATE and NEES per preset | `vio_pipeline` EuRoC runs | manual |
| E7 | Sanitizers on a real replay | ASan and UBSan clean end to end | run by hand | **missing** in CI |
| E8 | Latency | Real-time factor 1.0× or better (now 0.10×, p50 506 ms); synthetic budget 75 ms double, 50 ms float | `vio_latency.cpp` | gated, informational |
-| E9 | **Consistency gate:** told the true noise, the filter improves | One truth (a B-spline through the ground truth) feeds the synthesized IMU, the ideal tracks (1 and 0.5 px, noise in normalized coordinates) and the score; seeds 1–3, all 11 sequences, the same shipped S6d gate for every backend: no run with R matched worse than 10× OpenVINS mono; matched R beats the default R on the median; median (smallest world-yaw σ ÷ yaw σ at the gauge anchor) at least 0.5 | `scripts/euroc-consistency.sh --backend msckf` or `riekf` (results in `scripts/euroc-consistency.json`); `gt_spline.cpp` locks the truth (S2 closure, noise densities); `pose_consistency.cpp` the NEES split | **failing**, manual. NIS 1.00 on every sequence for both since . MSCKF: fails 1 (2 of 66 diverge) and 3 (yaw σ ratio 0.179, the S6 yaw leak), passes 2; R-IEKF: fails 1 only (the same 2 runs), passes 2 and 3 (0.984). Both divergences start from the shared initializer's seed (). Conditions 1 and 2 still move with the noise draw at three seeds (, ) |
+| E9 | **Consistency gate:** told the true noise, the filter improves | One truth (a B-spline through the ground truth) feeds the synthesized IMU, the ideal tracks (1 and 0.5 px, noise in normalized coordinates) and the score; seeds 1–3, all 11 sequences, the same shipped S6d gate for every backend: no run with R matched worse than 10× OpenVINS mono; matched R beats the default R on the median; median (smallest world-yaw σ ÷ yaw σ at the gauge anchor) at least 0.5 | `scripts/euroc-consistency.sh --backend msckf` or `riekf` (results in `scripts/euroc-consistency.json`); `gt_spline.cpp` locks the truth (S2 closure, noise densities); `pose_consistency.cpp` the NEES split | **failing**, manual. NIS 1.00 on every sequence for both since . MSCKF: fails 1 (2 of 66 diverge) and 3 (yaw σ ratio 0.152, the S6 yaw leak), passes 2; R-IEKF: fails 1 only (the same 2 runs), passes 2 and 3 (0.969). Both divergences are moving starts' velocity seeds, not the tilt seed (). Conditions 1 and 2 still move with the noise draw at three seeds (, ) |
## F. Engineering hygiene
diff --git a/docs/first-principles/signatures.mdx b/docs/first-principles/signatures.mdx
index 01534a7..896a7bb 100644
--- a/docs/first-principles/signatures.mdx
+++ b/docs/first-principles/signatures.mdx
@@ -1,6 +1,6 @@
---
title: Diagnostic Signatures and Pitfalls
-description: A symptom-to-cause table for VIO failures, each with the measurement that confirms it, and twenty-two pitfalls distilled from getting cortex's pipeline to work.
+description: A symptom-to-cause table for VIO failures, each with the measurement that confirms it, and twenty-five pitfalls distilled from getting cortex's pipeline to work.
---
import Issue from '@site/components/Issue.astro';
@@ -33,6 +33,9 @@ from its look-alikes; the [storyline](storyline.mdx) has the full account.
| A "consistent" control with NEES ≪ dof | Noise-free inputs: the control measures nothing | Inject noise matched to the filter's model | |
| Position variance under-predicted far from the origin | Process noise not mapped through the [p̂]× lever arm (invariant error) | Monte-Carlo of the propagated error about 20 m from the origin | |
| NIS ≈ 1 on slow sequences, 1.1–1.9 on fast ones, with R and Q exact | An IMU sample held over an interval it doesn't represent: a half-sample IMU–camera skew | Close S2 on a continuous truth at the real rate; average the interval's two samples and watch NIS return to 1.00 | |
+| Tilted from the start by about b_a / g, the tilt σ far smaller than the error | The seed took "up" from the accelerometer with b_a = 0 and its prior doesn't correlate tilt with bias | Tilt error and σ per frame against the truth (`vio_pipeline --out` writes `tilt.csv`); the seed's P_θb block | |
+| A tilt gap to the ground truth that no prior moves, fixed in the body frame | The ground truth's own attitude (and the bias its estimator fitted to it) | Rerun with a loose bias prior: the same answer means the data supports the filter; check the gap's frame | |
+| A moving start diverges, its tilt dragged off in the first second | A loose tilt prior meeting a wrong velocity seed: the updates explain the velocity by tilt | The seed's speed against the truth's; tighten nothing until the velocity prior is right | |
| Most new tracks die on their first frame | Detector margin narrower than the KLT pyramid footprint | Fraction of tracks lost on their first frame | |
## Pitfalls
@@ -86,3 +89,13 @@ from its look-alikes; the [storyline](storyline.mdx) has the full account.
([Aligning Sensor Samples](sample-alignment.mdx), ).
22. **Take every tolerance from what the next stage resolves, never from the current residual.** A
closure test fitted to what S2 achieved blessed the skew in item 21.
+23. **A prior is a joint statement.** At rest the accelerometer measures tilt and horizontal bias as
+ one sum; an isotropic prior claims they are independent and hands the sum to whichever σ is
+ larger. At rest only roll and pitch excursions separate them; with vision pinning the position,
+ turns and accelerations do too, and a level hover never does ().
+24. **Know which parts of the ground truth were measured.** EuRoC's bias and, in places, its
+ attitude are its estimator's outputs. V1_01's 0.55 m/s² bias came with a 2.5° body-frame
+ attitude offset that no prior moved; a target set against it can't be met.
+25. **An honest prior can still break a filter when another prior is wrong.** Loosening the
+ gravity-aligned seed's tilt to its true 15° let the updates absorb its wrong zero velocity, and
+ divergences tripled. Fix the wrong prior; don't loosen the right one to cover it.
diff --git a/docs/first-principles/storyline.mdx b/docs/first-principles/storyline.mdx
index a2284c5..3f58640 100644
--- a/docs/first-principles/storyline.mdx
+++ b/docs/first-principles/storyline.mdx
@@ -349,11 +349,57 @@ up to 624.
The stage contract is on [S5 — Feature Triangulation](../pipeline/stages/s5-triangulation.mdx).
+## Chapter 7. One quantity, seen once
+
+Both backends seeded roll and pitch from the accelerometer mean as if its bias were zero, and
+gave tilt and bias an isotropic, uncorrelated σ = 0.1. On V1_01 they stayed 2.4–2.6° tilted from
+the ground truth for the whole run and claimed 0.05° (roll/pitch ANEES 2,981,
+).
+
+- **The physics:** at rest the accelerometer reads f = Rᵀg·e_z + b_a. A seed that takes "up" from
+ f with b_a = 0 knows the tilt and the horizontal bias only as one sum: δθ⊥ = [û]ₓ b_a / g. A
+ tilt σ of 0.1 rad is 1 m/s² of gravity against 0.1 m/s² of bias, so the old prior put 99% of
+ the sum in tilt.
+- **The fix:** S1 now writes the block a static window implies, with the tilt and the bias
+ correlated (σ_tilt ≈ σ_ba / g to first order, about 1.2° for σ_ba = 0.2 m/s²), and seeds the
+ bias along up from |f| − g. A Monte Carlo of static seeds measured ANEES 4.93 against 5 (the test
+ asserts 4.5–5.5); on the synthetic world a
+ 0.5 m/s² bias is learned and the tilt falls under 0.01°.
+- **What it did not do:** V1_01 is still 2.4° off its ground truth. The same answer comes back
+ with σ_ba = 0.2, with 1.0 and with the old prior: b_a y ≈ 0.13 m/s², the value the filter finds
+ on all eleven sequences and the ground truth gives on the others. The gap is a fixed rotation in
+ the body frame. The ground truth's 0.55 m/s² bias is its own estimator's output, and a 2.5°
+ error in its attitude would produce exactly that bias. The data supports the filter, so the
+ issue's acceptance target (0.55 on V1_01) was set against a quantity the truth estimated, not
+ measured.
+- **What separates them is motion the camera sees.** At rest, turning about gravity leaves the
+ body's "up" where it was, and only roll and pitch excursions tell a tilt from a bias. With vision
+ pinning the position, a turn separates them too (a body-fixed bias turns with the body, a tilt
+ error stays fixed in the world), and so does an acceleration: the synthetic world learns its bias
+ with yaw-only rotation and a moving loop. A drone that hovers level keeps the sum.
+- **A loose prior that's honest can still break the filter.** A gravity-aligned seed (the
+ moving-start fallback) averages a window that accelerates: on the gate's MH_03 draws |f| − g is
+ 2.6 m/s², so the honest tilt σ is 15°. With it, three of the gate's five MH_03 draws diverged
+ against one (three of its 66 runs per backend, against two), because
+ that seed's zero velocity (σ 0.1 m/s) is the part that's wrong, and the first updates explained
+ the velocity by tilt. The gravity-aligned seed keeps the old prior until its velocity prior is
+ fixed.
+- **The two backends took it differently.** The MSCKF's worst roll/pitch NEES on the gate fell
+ from 24 to 4.8 and its rotation drift in the Vicon rooms 19–26%. The R-IEKF, which inherits the
+ seed, lost yaw σ on the V2 sequences (0.95 → 0.64) and drifted up to 26% more in rotation, a
+ bias-coupled path its invariant error doesn't close ().
+- **The issue's other diagnosis was wrong too.** Both backends diverge on the gate's MH_04 draw
+ from a dynamic seed, and #609 blamed its tilt. Measured, the seed's tilt was 0.14° off; its
+ speed was 0.05 m/s against the truth's 0.42. The divergence is the dynamic alignment's velocity
+ and scale, not this seed's tilt.
+
## What finishes the story
- The R-IEKF's NEES on real EuRoC () and its accuracy gap ().
- The time-offset state t_d, the one untested over-confidence candidate ().
-- The dynamic-initialization defects of Chapter 2.
+- The dynamic-initialization defects of Chapter 2, and the moving starts' velocity seeds of
+ Chapter 7 (the dynamic alignment's 0.05 m/s against 0.42 on the gate's MH_04; the
+ gravity-aligned seed's zero velocity).
- Real-time latency ().
- Why a front end without border churn costs about 10% EuRoC ATE (): the
detection margin that fixes it ships off until that is understood.
diff --git a/docs/methodology/euroc-consistency-gate.mdx b/docs/methodology/euroc-consistency-gate.mdx
index 1eaf956..051ecc9 100644
--- a/docs/methodology/euroc-consistency-gate.mdx
+++ b/docs/methodology/euroc-consistency-gate.mdx
@@ -112,7 +112,7 @@ on grid complete:
| every sequence's matched runs have normalized NIS within the χ² band for their dof | `scripts/euroc-consistency.sh` | holds — 1.00 on every sequence for both backends at f1551cc. Before #616 it was 1.1–1.9 on V1_02, V1_03, V2_02, V2_03: S2's sample-hold skew, an estimator defect |
| the shipped S2, fed the synthesized IMU noise-free, closes on the truth within what the camera resolves (a tenth of 0.5 px over a 0.55 s clone window: 1.1e-4 rad, 0.22 mm at 2 m), at second order (bound 1.09e-4 rad rotation, 0.22 mm position at 2 m, over a 0.55 s clone window, from what **run_filter** resolves: the backend's update resolves a clone pose only to the track noise, 0.5 px at fx = 458.654 (1.09e-3 rad); propagation error must stay a tenth of that over the clone window so it never competes with the measurement, and the same angle at a 2 m landmark depth is 0.22 mm) | `tests/tools/gt_spline.cpp` “gt spline: S2 closure, the shipped S2 within what the camera resolves…” | holds — since #616, S2 integrates each interval from both samples by RK4: 0.27 of the rotation bound and 0.10 of the position bound at 200 Hz on a 1.3 rad/s roll/pitch flight, converging at second order. Before it, the end-sample hold missed the rotation bound 28× at first order; the companion tests pin that, and that centring the input fixes the rotation while the integrator fixes the position |
| every backend runs with the same S6d gate line | `scripts/euroc-consistency.sh` | holds — since #608 both use the χ² quantile; before it the R-IEKF used 16 per dof |
-| each condition's result changes with the estimator, not with the gate's line | — | holds — at c6b87ec each remaining failure was rerun with the gate at 5 per dof: one divergence per backend vanished, the others persisted, and condition 2's fast-sequence loss was #616's S2 skew. At f1551cc condition 2 passes for both; both backends diverge on the same two noise draws, from the shared initializer's seed (#609); the MSCKF's condition 3 is S6's yaw leak. Conditions 1 and 2 still depend on the noise draw at three seeds (#613) |
+| each condition's result changes with the estimator, not with the gate's line | — | holds — at c6b87ec each remaining failure was rerun with the gate at 5 per dof: one divergence per backend vanished, the others persisted, and condition 2's fast-sequence loss was #616's S2 skew. At f1551cc condition 2 passes for both; both backends diverge on the same two noise draws, which #609 traced to the moving starts' velocity seeds (MH_04's dynamic seed 0.05 m/s against the truth's 0.42; MH_03's gravity-aligned zero velocity), not the tilt; the MSCKF's condition 3 is S6's yaw leak. Conditions 1 and 2 still depend on the noise draw at three seeds (#613) |
## Gap report
diff --git a/docs/methodology/openvins-msckf.mdx b/docs/methodology/openvins-msckf.mdx
index 49a6bc5..41d1e30 100644
--- a/docs/methodology/openvins-msckf.mdx
+++ b/docs/methodology/openvins-msckf.mdx
@@ -66,8 +66,16 @@ Each has the shape *(state, inputs, params) → (state′, diagnostics)*.
- *Pre:* a successful seed
- *Post:* P₀ symmetric positive-definite, in the error convention every later node assumes
- *Test:* `tests/sdk/msckf_backend.cpp` “the backend initializes attitude from a stationary window…”
-- *Decision — the prior P₀:* isotropic σ = 0.1 on every core state over the tilt–accelerometer-bias correlation an IMU window implies (#609). **no reason recorded**
-- *Deviates from the paper:* the prior is an isotropic σ = 0.1 with no tilt-bias correlation, whatever the init method (#609)
+- *Test:* `tests/sdk/tilt_bias_prior.cpp` “S1 prior: P0 is the closed-form tilt-bias block about the seed's up…”
+- *Test:* `tests/sdk/tilt_bias_prior.cpp` “S1 prior: P0 covers a static seed's tilt and bias error (NEES 5)…”
+- *Test:* `tests/sdk/tilt_bias_prior.cpp` “MSCKF backend: a 0.5 m/s^2 accelerometer bias is learned once the motion rotates…”
+- *Decision — the attitude and accelerometer-bias blocks of P₀ on a static seed:* the block the window implies (#609): P_θθ = (σ_ba² + σ_f²)/g²·(I − ûûᵀ) + σ_ψ²ûûᵀ, P_θb = σ_ba²/g·[û]ₓ, P_bb = σ_ba²(I − ûûᵀ) + σ_∥²ûûᵀ (stages::s1_initialization::TiltBiasPrior) over isotropic σ = 0.1 on every core state, uncorrelated (to #609), a tilt σ of σ_ba/g without the correlation. the window measures f̄ = g·u + b_a once, so a seed that takes û from f̄ with b_a = 0 knows the tilt and the horizontal bias only as one sum, δθ⊥ = [û]ₓ(b_a + n̄)/g. The isotropic prior put 99% of that sum in tilt (0.1 rad of tilt is 1 m/s² of gravity, against 0.1 m/s² of bias), so the filter never learned EuRoC's bias and stayed tilted 2.4–2.6° on V1_01. Without the correlation the prior would claim the tilt and bias independent, which the seed is not
+- *Decision — σ_ba, the accelerometer-bias prior:* 0.2 m/s² per axis (VioConfig::accel_bias_prior_sigma) over 0.1 (the isotropic value to #609), a datasheet turn-on bias. EuRoC's ground-truth bias reaches 0.55 m/s² (V1_01 y) and is 0.09–0.20 elsewhere; 0.2 puts the largest inside 3σ. The tilt σ that follows is 0.2/g = 1.2°
+- *Decision — σ_f, the error of the window's mean specific force:* the larger of the sample scatter over √N and σ_a/√(window) over the white-noise model alone, zero. it bounds how well the window resolves the tilt–bias sum. EuRoC's rotor vibration gives 0.03–0.28 m/s² of per-sample scatter, far above the white-noise model's 0.028
+- *Decision — the prior of a gravity-aligned seed (the moving-start fallback):* the fresh state's isotropic σ = 0.1 over the static block with σ_f² adding the window's per-sample scatter and (\|f̄\| − g)², the static block with σ_f² adding (\|f̄\| − g)² only. measured on the gate's five MH_03 draws (four gravity-aligned): a moving window's mean force includes its mean acceleration, \|f̄\| − g = 2.6 m/s² there, so either block gives a 15–16° tilt σ, and the seed's zero velocity (σ 0.1 m/s) is then the wrong part. The first updates explain the velocity by tilt (2.4° → 12–19°), and three draws diverge per backend (3.9–5.2 km at 0.5 px seed 1) against one with the isotropic 5.7°. On the real MH_03 the block gained 0.500 → 0.452 m. The defect is the zero-velocity seed of a moving start
+- *Decision — the accelerometer bias along up at a static seed:* seeded from \|f̄\| − g, with σ_∥² = σ_f² + 0.026², half the spread of surface gravity over zero, with σ_ba (to #609). with g known the window measures the bias along up. On EuRoC \|f̄\| − 9.81 is −0.011 to −0.035 m/s² at each sequence's stillest early second against the ground truth's −0.024 to −0.058 along up: it halves that component's error, and the residual is within the gravity σ
+- *Decision — the prior of a dynamic (visual-inertial alignment) seed:* the fresh state's isotropic σ = 0.1, the alignment's bias estimate kept in the diagnostics, not seeded over the static block about the alignment's gravity direction, the alignment's bias estimate and its least-squares covariance (#609's scope). measured: the dynamic seed's tilt is accurate (0.14° on the gate's MH_04 draw that diverges, 1.0–1.4° on EuRoC MH_01/MH_02, inside the isotropic 5.7°), and its tilt comes from the alignment, not from the accelerometer mean, so the static correlation does not apply. The alignment's solve is unweighted with a regularized bias, so H⁻¹ is not a covariance. The MH_04 gate divergence is the seeded velocity (0.05 m/s against the truth's 0.42), not the tilt
+- *Deviates from the paper:* the S1 bench seeds the mean with the isotropic P₀: its windows carry no duration for the #609 prior, which tests/sdk/tilt_bias_prior.cpp checks
**propagate** (S2, IMU propagation)
@@ -136,14 +144,13 @@ Each has the shape *(state, inputs, params) → (state′, diagnostics)*.
| **CameraCalib** | input | intrinsics (fx, fy, cx, cy), radtan distortion, camera↔IMU extrinsic (R_imu_cam, p_imu_cam) | `math/include/branes/math/cameras/pinhole_radtan.hpp` `PinholeRadtanCamera` | ✓ R_imu_cam ∈ SO(3); ✓ intrinsics positive, principal point inside the image |
| **NormalizedObs** | data | feature id, camera id, normalized image point (x, y) | `sdk/include/branes/sdk/msckf/stages/s0_sensor_model.hpp` `Result` | ✗ finite, and the ray is in front of the camera (z > 0) |
| **TrackDB** | data | feature id → [(clone time, normalized observation)], plus the tracks that ended this frame | `sdk/include/branes/sdk/msckf/stages/s4_frontend.hpp` `TrackTable` | ✗ every observation's clone time is a live clone, or the observation is purged; ✓ points inside the image; ✓ track length at most the clone window |
-| **NavSeed** | data | R, v, p, b_g, b_a, the time, and the prior P₀ in the filter's error convention | `sdk/include/branes/sdk/msckf/stages/s1_initialization.hpp` `seed_from_imu` | ✓ gravity sign and magnitude; ✓ P₀ symmetric positive-definite; ✗ P₀ states what the init window does and does not determine (tilt-bias correlation) |
+| **NavSeed** | data | R, v, p, b_g, b_a, the time, and the prior P₀ in the filter's error convention | `sdk/include/branes/sdk/msckf/stages/s1_initialization.hpp` `seed_from_imu` | ✓ gravity sign and magnitude; ✓ P₀ symmetric positive-definite; ✓ P₀ states what the init window does and does not determine (tilt-bias correlation) |
| **FilterState** | state | nav (θ, p, v, b_g, b_a: 15), clones (θ, p: 6 each), optional camera↔IMU extrinsics (6 per camera), the joint covariance P, and the error-state convention every Jacobian assumes | `sdk/include/branes/sdk/msckf/state.hpp` `State` | ✓ P symmetric; ✓ P positive semidefinite; ✓ R and every clone rotation ∈ SO(3); ✗ the layout and the error convention agree with every node's Jacobians |
| **Landmark** | data | feature position p_f (world), the triangulation's parallax and conditioning | `sdk/include/branes/sdk/msckf/stages/s5_triangulation.hpp` `Result` | ✓ positive depth in every observing camera; ✓ normal-matrix condition number reported |
| **UpdateDiag** | output | NIS, dof, the gate verdict, and where a rejected track stopped | `sdk/include/branes/sdk/msckf/stages/s6_msckf_update.hpp` `Result` | ✓ NIS ~ χ²(dof) over windows (bound χ²₁₋α(dof) per update, α = 0.05, from what **msckf_update** resolves: the S6d gate rejects above it, so a consistent filter loses α of its inliers and no more) |
- **ImageFrame** precision: integer pixels; no arithmetic-type choice
- **NormalizedObs** precision: the narrowest edge that keeps sub-pixel accuracy; float holds (#532)
-- **NavSeed** deviates from the paper: P₀ is an isotropic σ = 0.1 on every core state, with no tilt-bias correlation (#609)
- **FilterState** precision: stored in double; the P-writing stages must stay double (#532)
- **FilterState** deviates from the paper: error convention R = R̂·Exp(δθ) (body-frame right perturbation); OpenVINS uses a left quaternion error. Legitimate if consistent everywhere.
@@ -179,33 +186,30 @@ on image (20 Hz):
| NIS ≈ dim(measurement) per update | `tests/sdk/consistency.cpp` “NEES/NIS flag over- and under-confidence…” | unknown |
| innovations zero-mean and white | `tests/sdk/innovation_whiteness.cpp` “i.i.d. N(0,1) innovations read as zero-mean and white…” | unknown |
| no information gained along global yaw and translation (the 4-D unobservable space) | `tests/tools/observability_ledger.cpp` “Observability ledger: on the synthetic world the leak is yaw…” | failing — the yaw leak (#480); fixed structurally by FEJ (#606) or the R-IEKF (#608) |
-| told the true noise, the filter improves and its yaw σ never collapses (EuRoC ideal tracks) | `scripts/euroc-consistency.sh` | failing — with one truth and S2 integrating both samples (#616, median NIS 1.00): fails 1 (2 of 66 at 1 px, MH_03 and MH_04, the same draws as the R-IEKF: the shared tilt and bias seed, #609), passes 2 (0.45× vs 0.52×; 0.33× vs 0.53×), fails 3 (yaw σ ratio 0.179: the S6 yaw leak, #480) |
+| told the true noise, the filter improves and its yaw σ never collapses (EuRoC ideal tracks) | `scripts/euroc-consistency.sh` | failing — with one truth, S2 integrating both samples (#616) and the tilt–bias seed prior (#609), median NIS 1.00: fails 1 (2 of 66 at 1 px, MH_03 and MH_04, the same draws as the R-IEKF: moving starts' velocity seeds, not the tilt), passes 2 (0.36× vs 0.64×; 0.25× vs 0.58×), fails 3 (yaw σ ratio 0.152: the S6 yaw leak, #480) |
| 4-DoF ATE and rotation RPE within a factor of OpenVINS mono on the 11 EuRoC sequences | `scripts/euroc-baseline.sh` | failing — median 2.8× OpenVINS mono (#595) |
## Gap report
-What `pipeline_graph.py gaps` reports for this spec (23 items):
+What `pipeline_graph.py gaps` reports for this spec (20 items):
- edge 'ImageFrame': no validator for "t increasing and inside the IMU span"
- edge 'ImageFrame': no validator for "the image matches the calibration's resolution"
- edge 'NormalizedObs': no validator for "finite, and the ray is in front of the camera (z > 0)"
- edge 'TrackDB': no validator for "every observation's clone time is a live clone, or the observation is purged"
-- edge 'NavSeed': no validator for "P₀ states what the init window does and does not determine (tilt-bias correlation)"
-- edge 'NavSeed': deviates from the paper: P₀ is an isotropic σ = 0.1 on every core state, with no tilt-bias correlation (#609)
- edge 'FilterState': no validator for "the layout and the error convention agree with every node's Jacobians"
- edge 'FilterState': deviates from the paper: error convention R = R̂·Exp(δθ) (body-frame right perturbation); OpenVINS uses a left quaternion error. Legitimate if consistent everywhere.
- node 'track': deviates from the paper: no RANSAC outlier rejection in the front end; the S6d gate carries it
- node 'track': deviates from the paper: the shipped default churns (>50% of new tracks lost on their first frame); the border margin is opt-in (#474, #574)
- node 'initialize': deviates from the paper: the dynamic path is excitation-limited on EuRoC and has open SfM defects (#254, #499–#501)
-- node 'seed_state': deviates from the paper: the prior is an isotropic σ = 0.1 with no tilt-bias correlation, whatever the init method (#609)
+- node 'seed_state': deviates from the paper: the S1 bench seeds the mean with the isotropic P₀: its windows carry no duration for the #609 prior, which tests/sdk/tilt_bias_prior.cpp checks
- node 'propagate': deviates from the paper: no first-estimates Jacobians on Φ (FEJ, #606)
- node 'msckf_update': deviates from the paper: no first-estimates Jacobians on H (FEJ, #606); the yaw leak is S6 moving the mean (95–99%, #480)
- node 'msckf_update': deviates from the paper: S6c compression is the identity (one feature per update leaves 2m − 3 rows)
- node 'slam_update' (S7): not implemented
- node 'online_calib' (S10): partially implemented
- node 'online_calib': deviates from the paper: extrinsics only (VioConfig::estimate_extrinsics, off by default); no intrinsics; no time offset (#332)
-- node 'seed_state': decision 'the prior P₀' has no recorded reason (chosen: isotropic σ = 0.1 on every core state)
- system invariant 'nees': failing — the body-frame filter is over-confident on EuRoC (#212, #605)
- system invariant 'gauge': failing — the yaw leak (#480); fixed structurally by FEJ (#606) or the R-IEKF (#608)
-- system invariant 'consistency_gate': failing — with one truth and S2 integrating both samples (#616, median NIS 1.00): fails 1 (2 of 66 at 1 px, MH_03 and MH_04, the same draws as the R-IEKF: the shared tilt and bias seed, #609), passes 2 (0.45× vs 0.52×; 0.33× vs 0.53×), fails 3 (yaw σ ratio 0.179: the S6 yaw leak, #480)
+- system invariant 'consistency_gate': failing — with one truth, S2 integrating both samples (#616) and the tilt–bias seed prior (#609), median NIS 1.00: fails 1 (2 of 66 at 1 px, MH_03 and MH_04, the same draws as the R-IEKF: moving starts' velocity seeds, not the tilt), passes 2 (0.36× vs 0.64×; 0.25× vs 0.58×), fails 3 (yaw σ ratio 0.152: the S6 yaw leak, #480)
- system invariant 'accuracy': failing — median 2.8× OpenVINS mono (#595)
diff --git a/docs/methodology/riekf-msckf.mdx b/docs/methodology/riekf-msckf.mdx
index 67afaae..8c16a2c 100644
--- a/docs/methodology/riekf-msckf.mdx
+++ b/docs/methodology/riekf-msckf.mdx
@@ -71,7 +71,7 @@ Each has the shape *(state, inputs, params) → (state′, diagnostics)*.
- *Post:* the mean equals the seed; P₀ = J·P_body·Jᵀ, positive-definite; the IMU sample in effect at the seed time is held, so propagation resumes at the seed time
- *Test:* `tests/tools/invariant_backend_adapter.cpp` “invariant adapter: the body-to-invariant map matches the exact error…”
- *Test:* `tests/tools/invariant_backend_adapter.cpp` “invariant adapter: initializes through the shipped MSCKF with its seed…”
-- *Deviates from the paper:* the base's prior is itself an isotropic σ = 0.1 with no tilt-bias correlation (#609); this node maps it faithfully
+- *Test:* `tests/tools/invariant_backend_adapter.cpp` “invariant adapter: the seed's tilt-bias correlation reaches the invariant prior…”
**propagate** (S2, Invariant IMU propagation)
@@ -128,14 +128,13 @@ Each has the shape *(state, inputs, params) → (state′, diagnostics)*.
| **CameraCalib** | input | intrinsics (fx, fy, cx, cy), radtan distortion, camera↔IMU extrinsic (R_imu_cam, p_imu_cam) | `math/include/branes/math/cameras/pinhole_radtan.hpp` `PinholeRadtanCamera` | ✓ R_imu_cam ∈ SO(3); ✓ intrinsics positive, principal point inside the image |
| **NormalizedObs** | data | feature id, camera id, normalized image point (x, y) | `sdk/include/branes/sdk/msckf/stages/s0_sensor_model.hpp` `Result` | ✗ finite, and the ray is in front of the camera (z > 0) |
| **TrackDB** | data | feature id → [(clone time, normalized observation)], plus the tracks that ended this frame | `sdk/include/branes/sdk/msckf/stages/s4_frontend.hpp` `TrackTable` | ✗ every observation's clone time is a live clone, or the observation is purged; ✓ points inside the image; ✓ track length at most the clone window |
-| **NavSeed** | data | R, v, p, b_g, b_a, the time, and the prior P₀ in the filter's error convention | `sdk/include/branes/sdk/msckf/stages/s1_initialization.hpp` `seed_from_imu` | ✓ gravity sign and magnitude; ✓ P₀ symmetric positive-definite; ✗ P₀ states what the init window does and does not determine (tilt-bias correlation) |
+| **NavSeed** | data | R, v, p, b_g, b_a, the time, and the prior P₀ in the filter's error convention | `sdk/include/branes/sdk/msckf/stages/s1_initialization.hpp` `seed_from_imu` | ✓ gravity sign and magnitude; ✓ P₀ symmetric positive-definite; ✓ P₀ states what the init window does and does not determine (tilt-bias correlation) |
| **FilterState** | state | nav on SE₂(3) with additive biases (θ, ν, ρ, b_g, b_a: 15), clones (θ, ρ: 6 each) in the same right-invariant error, the joint covariance P | `sdk/include/branes/sdk/msckf/invariant_propagator.hpp` `InvariantNavState` | ✓ P symmetric; ✓ P positive semidefinite; ✓ R and every clone rotation ∈ SO(3); ✗ every Jacobian and retraction uses R = Exp(δθ)·R̂, v = Exp(δθ)·v̂ + ν, p = Exp(δθ)·p̂ + ρ |
| **Landmark** | data | feature position p_f (world), the triangulation's parallax and conditioning | `sdk/include/branes/sdk/msckf/stages/s5_triangulation.hpp` `Result` | ✓ positive depth in every observing camera; ✓ normal-matrix condition number reported |
| **UpdateDiag** | output | NIS, dof, the gate verdict, and where a rejected track stopped | `sdk/include/branes/sdk/msckf/stages/s6_msckf_update.hpp` `Result` | ✓ NIS ~ χ²(dof) over windows (bound χ²₁₋α(dof) per update, α = 0.05, from what **msckf_update** resolves: the S6d gate rejects above it, so a consistent filter loses α of its inliers and no more) |
- **ImageFrame** precision: integer pixels; no arithmetic-type choice
- **NormalizedObs** precision: the narrowest edge that keeps sub-pixel accuracy; float holds (#532)
-- **NavSeed** deviates from the paper: P₀ is an isotropic σ = 0.1 on every core state, with no tilt-bias correlation (#609)
- **FilterState** precision: stored in double; the Joseph form on EuRoC (the square-root form diverges there)
## Schedule
@@ -169,13 +168,13 @@ on image (20 Hz):
| NIS ≈ dim(measurement) per update | `sdk/include/branes/sdk/msckf/invariant_vio_backend.hpp` `nis_consistency` | failing — real KLT tracks: normalized NIS 0.97–2.55 on MH, V1_01 and V2_01; 6.8–15.4 on V1_02, V1_03, V2_02 and V2_03 (#608) |
| innovations zero-mean and white | `sdk/include/branes/sdk/msckf/invariant_vio_backend.hpp` `innovation_whiteness` | unknown — reported since #608 |
| no information gained along global yaw and translation (the 4-D unobservable space) | `tests/sdk/msckf_invariant_backend.cpp` “invariant backend: the marginalized joint update annihilates the 4-DoF gauge…” | holds |
-| told the true noise, the filter improves and its yaw σ never collapses (EuRoC ideal tracks) | `scripts/euroc-consistency.sh` | failing — with one truth and S2 integrating both samples (#616, median NIS 1.00): fails 1 (2 of 66 at 1 px, MH_03 and MH_04, the same draws as the MSCKF: the shared tilt and bias seed, #609), passes 2 (0.40× vs 0.41×; 0.18× vs 0.36×), passes 3 (0.984) |
+| told the true noise, the filter improves and its yaw σ never collapses (EuRoC ideal tracks) | `scripts/euroc-consistency.sh` | failing — with one truth, S2 integrating both samples (#616) and the tilt–bias seed prior (#609), median NIS 1.00: fails 1 (2 of 66 at 1 px, MH_03 and MH_04, the same draws as the MSCKF: moving starts' velocity seeds, not the tilt), passes 2 (0.32× vs 0.53×; 0.19× vs 0.40×), passes 3 (0.969; V2_01 and V2_02 fall to 0.64–0.67 at 0.5 px, #620) |
| 4-DoF ATE and rotation RPE within a factor of OpenVINS mono on the 11 EuRoC sequences | `scripts/euroc-baseline.sh` | failing — median 2.62× OpenVINS mono (the MSCKF 2.83×; 7 of 11 sequences better); rotation RPE 0.90° vs 2.21° (#608) |
## Changes from the base
- replace edge **FilterState**: the error convention changes; every consumer is revisited below
-- replace node **seed_state**: #608
+- replace node **seed_state**: #608; since #609 the base's prior correlates tilt and accelerometer bias, and J carries the correlation (δθ_world = R̂·δθ_body, b_a unchanged)
- replace node **propagate**
- replace node **augment**
- replace node **triangulate**
@@ -187,20 +186,17 @@ on image (20 Hz):
## Gap report
-What `pipeline_graph.py gaps` reports for this spec (22 items):
+What `pipeline_graph.py gaps` reports for this spec (19 items):
- edge 'ImageFrame': no validator for "t increasing and inside the IMU span"
- edge 'ImageFrame': no validator for "the image matches the calibration's resolution"
- edge 'NormalizedObs': no validator for "finite, and the ray is in front of the camera (z > 0)"
- edge 'TrackDB': no validator for "every observation's clone time is a live clone, or the observation is purged"
-- edge 'NavSeed': no validator for "P₀ states what the init window does and does not determine (tilt-bias correlation)"
-- edge 'NavSeed': deviates from the paper: P₀ is an isotropic σ = 0.1 on every core state, with no tilt-bias correlation (#609)
- edge 'FilterState': no validator for "every Jacobian and retraction uses R = Exp(δθ)·R̂, v = Exp(δθ)·v̂ + ν, p = Exp(δθ)·p̂ + ρ"
- node 'track': deviates from the paper: no RANSAC outlier rejection in the front end; the S6d gate carries it
- node 'track': deviates from the paper: the shipped default churns (>50% of new tracks lost on their first frame); the border margin is opt-in (#474, #574)
- node 'initialize': deviates from the paper: the dynamic path is excitation-limited on EuRoC and has open SfM defects (#254, #499–#501)
- node 'seed_state': no stage bench
-- node 'seed_state': deviates from the paper: the base's prior is itself an isotropic σ = 0.1 with no tilt-bias correlation (#609); this node maps it faithfully
- node 'augment': no stage bench
- node 'triangulate': no stage bench
- node 'msckf_update': no stage bench
@@ -209,5 +205,5 @@ What `pipeline_graph.py gaps` reports for this spec (22 items):
- node 'slam_update' (S7): not implemented
- node 'online_calib' (S10): not implemented
- system invariant 'nis': failing — real KLT tracks: normalized NIS 0.97–2.55 on MH, V1_01 and V2_01; 6.8–15.4 on V1_02, V1_03, V2_02 and V2_03 (#608)
-- system invariant 'consistency_gate': failing — with one truth and S2 integrating both samples (#616, median NIS 1.00): fails 1 (2 of 66 at 1 px, MH_03 and MH_04, the same draws as the MSCKF: the shared tilt and bias seed, #609), passes 2 (0.40× vs 0.41×; 0.18× vs 0.36×), passes 3 (0.984)
+- system invariant 'consistency_gate': failing — with one truth, S2 integrating both samples (#616) and the tilt–bias seed prior (#609), median NIS 1.00: fails 1 (2 of 66 at 1 px, MH_03 and MH_04, the same draws as the MSCKF: moving starts' velocity seeds, not the tilt), passes 2 (0.32× vs 0.53×; 0.19× vs 0.40×), passes 3 (0.969; V2_01 and V2_02 fall to 0.64–0.67 at 0.5 px, #620)
- system invariant 'accuracy': failing — median 2.62× OpenVINS mono (the MSCKF 2.83×; 7 of 11 sequences better); rotation RPE 0.90° vs 2.21° (#608)
diff --git a/docs/methodology/specs/euroc-consistency-gate.yaml b/docs/methodology/specs/euroc-consistency-gate.yaml
index 5859692..01cae04 100644
--- a/docs/methodology/specs/euroc-consistency-gate.yaml
+++ b/docs/methodology/specs/euroc-consistency-gate.yaml
@@ -274,4 +274,4 @@ system_invariants:
- id: verdict_attributable
text: "each condition's result changes with the estimator, not with the gate's line"
status: holds
- note: "at c6b87ec each remaining failure was rerun with the gate at 5 per dof: one divergence per backend vanished, the others persisted, and condition 2's fast-sequence loss was #616's S2 skew. At f1551cc condition 2 passes for both; both backends diverge on the same two noise draws, from the shared initializer's seed (#609); the MSCKF's condition 3 is S6's yaw leak. Conditions 1 and 2 still depend on the noise draw at three seeds (#613)"
+ note: "at c6b87ec each remaining failure was rerun with the gate at 5 per dof: one divergence per backend vanished, the others persisted, and condition 2's fast-sequence loss was #616's S2 skew. At f1551cc condition 2 passes for both; both backends diverge on the same two noise draws, which #609 traced to the moving starts' velocity seeds (MH_04's dynamic seed 0.05 m/s against the truth's 0.42; MH_03's gravity-aligned zero velocity), not the tilt; the MSCKF's condition 3 is S6's yaw leak. Conditions 1 and 2 still depend on the noise draw at three seeds (#613)"
diff --git a/docs/methodology/specs/openvins-msckf.yaml b/docs/methodology/specs/openvins-msckf.yaml
index 34fa96f..a074ab1 100644
--- a/docs/methodology/specs/openvins-msckf.yaml
+++ b/docs/methodology/specs/openvins-msckf.yaml
@@ -99,8 +99,7 @@ edges:
- text: P₀ symmetric positive-definite
check: {file: sdk/include/branes/sdk/eval/invariants/matrix_checks.hpp, symbol: check_spd}
- text: P₀ states what the init window does and does not determine (tilt-bias correlation)
- deviations:
- - P₀ is an isotropic σ = 0.1 on every core state, with no tilt-bias correlation (#609)
+ check: {file: tests/sdk/tilt_bias_prior.cpp, case: "S1 prior: P0 covers a static seed's tilt and bias error (NEES 5)"}
- id: FilterState
kind: state
@@ -227,15 +226,39 @@ nodes:
- "P₀ symmetric positive-definite, in the error convention every later node assumes"
code: sdk/include/branes/sdk/msckf/stages/s1_initialization.hpp
bench: {file: tests/tools/stage_bench_s0_s1.cpp, case: "S1 bench: static paths pass every fixture"}
+ deviations:
+ - "the S1 bench seeds the mean with the isotropic P₀: its windows carry no duration for the #609 prior, which tests/sdk/tilt_bias_prior.cpp checks"
tests:
- {file: tests/sdk/msckf_backend.cpp, case: "the backend initializes attitude from a stationary window"}
+ - {file: tests/sdk/tilt_bias_prior.cpp, case: "S1 prior: P0 is the closed-form tilt-bias block about the seed's up"}
+ - {file: tests/sdk/tilt_bias_prior.cpp, case: "S1 prior: P0 covers a static seed's tilt and bias error (NEES 5)"}
+ - {file: tests/sdk/tilt_bias_prior.cpp, case: "MSCKF backend: a 0.5 m/s^2 accelerometer bias is learned once the motion rotates"}
status: implemented
- deviations:
- - the prior is an isotropic σ = 0.1 with no tilt-bias correlation, whatever the init method (#609)
decisions:
- - question: the prior P₀
- chosen: "isotropic σ = 0.1 on every core state"
- alternatives: ["the tilt–accelerometer-bias correlation an IMU window implies (#609)"]
+ - question: the attitude and accelerometer-bias blocks of P₀ on a static seed
+ chosen: "the block the window implies (#609): P_θθ = (σ_ba² + σ_f²)/g²·(I − ûûᵀ) + σ_ψ²ûûᵀ, P_θb = σ_ba²/g·[û]ₓ, P_bb = σ_ba²(I − ûûᵀ) + σ_∥²ûûᵀ (stages::s1_initialization::TiltBiasPrior)"
+ alternatives: ["isotropic σ = 0.1 on every core state, uncorrelated (to #609)", "a tilt σ of σ_ba/g without the correlation"]
+ reason: "the window measures f̄ = g·u + b_a once, so a seed that takes û from f̄ with b_a = 0 knows the tilt and the horizontal bias only as one sum, δθ⊥ = [û]ₓ(b_a + n̄)/g. The isotropic prior put 99% of that sum in tilt (0.1 rad of tilt is 1 m/s² of gravity, against 0.1 m/s² of bias), so the filter never learned EuRoC's bias and stayed tilted 2.4–2.6° on V1_01. Without the correlation the prior would claim the tilt and bias independent, which the seed is not"
+ - question: σ_ba, the accelerometer-bias prior
+ chosen: "0.2 m/s² per axis (VioConfig::accel_bias_prior_sigma)"
+ alternatives: ["0.1 (the isotropic value to #609)", "a datasheet turn-on bias"]
+ reason: "EuRoC's ground-truth bias reaches 0.55 m/s² (V1_01 y) and is 0.09–0.20 elsewhere; 0.2 puts the largest inside 3σ. The tilt σ that follows is 0.2/g = 1.2°"
+ - question: σ_f, the error of the window's mean specific force
+ chosen: "the larger of the sample scatter over √N and σ_a/√(window)"
+ alternatives: ["the white-noise model alone", "zero"]
+ reason: "it bounds how well the window resolves the tilt–bias sum. EuRoC's rotor vibration gives 0.03–0.28 m/s² of per-sample scatter, far above the white-noise model's 0.028"
+ - question: the prior of a gravity-aligned seed (the moving-start fallback)
+ chosen: "the fresh state's isotropic σ = 0.1"
+ alternatives: ["the static block with σ_f² adding the window's per-sample scatter and (|f̄| − g)²", "the static block with σ_f² adding (|f̄| − g)² only"]
+ reason: "measured on the gate's five MH_03 draws (four gravity-aligned): a moving window's mean force includes its mean acceleration, |f̄| − g = 2.6 m/s² there, so either block gives a 15–16° tilt σ, and the seed's zero velocity (σ 0.1 m/s) is then the wrong part. The first updates explain the velocity by tilt (2.4° → 12–19°), and three draws diverge per backend (3.9–5.2 km at 0.5 px seed 1) against one with the isotropic 5.7°. On the real MH_03 the block gained 0.500 → 0.452 m. The defect is the zero-velocity seed of a moving start"
+ - question: the accelerometer bias along up at a static seed
+ chosen: "seeded from |f̄| − g, with σ_∥² = σ_f² + 0.026², half the spread of surface gravity"
+ alternatives: ["zero, with σ_ba (to #609)"]
+ reason: "with g known the window measures the bias along up. On EuRoC |f̄| − 9.81 is −0.011 to −0.035 m/s² at each sequence's stillest early second against the ground truth's −0.024 to −0.058 along up: it halves that component's error, and the residual is within the gravity σ"
+ - question: the prior of a dynamic (visual-inertial alignment) seed
+ chosen: "the fresh state's isotropic σ = 0.1, the alignment's bias estimate kept in the diagnostics, not seeded"
+ alternatives: ["the static block about the alignment's gravity direction", "the alignment's bias estimate and its least-squares covariance (#609's scope)"]
+ reason: "measured: the dynamic seed's tilt is accurate (0.14° on the gate's MH_04 draw that diverges, 1.0–1.4° on EuRoC MH_01/MH_02, inside the isotropic 5.7°), and its tilt comes from the alignment, not from the accelerometer mean, so the static correlation does not apply. The alignment's solve is unweighted with a regularized bias, so H⁻¹ is not a covariance. The MH_04 gate divergence is the seeded velocity (0.05 m/s against the truth's 0.42), not the tilt"
- id: propagate
stage: S2
@@ -475,7 +498,7 @@ system_invariants:
text: told the true noise, the filter improves and its yaw σ never collapses (EuRoC ideal tracks)
check: {file: scripts/euroc-consistency.sh}
status: failing
- note: "with one truth and S2 integrating both samples (#616, median NIS 1.00): fails 1 (2 of 66 at 1 px, MH_03 and MH_04, the same draws as the R-IEKF: the shared tilt and bias seed, #609), passes 2 (0.45× vs 0.52×; 0.33× vs 0.53×), fails 3 (yaw σ ratio 0.179: the S6 yaw leak, #480)"
+ note: "with one truth, S2 integrating both samples (#616) and the tilt–bias seed prior (#609), median NIS 1.00: fails 1 (2 of 66 at 1 px, MH_03 and MH_04, the same draws as the R-IEKF: moving starts' velocity seeds, not the tilt), passes 2 (0.36× vs 0.64×; 0.25× vs 0.58×), fails 3 (yaw σ ratio 0.152: the S6 yaw leak, #480)"
- id: accuracy
text: 4-DoF ATE and rotation RPE within a factor of OpenVINS mono on the 11 EuRoC sequences
check: {file: scripts/euroc-baseline.sh}
diff --git a/docs/methodology/specs/riekf-msckf.yaml b/docs/methodology/specs/riekf-msckf.yaml
index cdb24ed..3965e53 100644
--- a/docs/methodology/specs/riekf-msckf.yaml
+++ b/docs/methodology/specs/riekf-msckf.yaml
@@ -74,10 +74,9 @@ nodes:
tests:
- {file: tests/tools/invariant_backend_adapter.cpp, case: "invariant adapter: the body-to-invariant map matches the exact error"}
- {file: tests/tools/invariant_backend_adapter.cpp, case: "invariant adapter: initializes through the shipped MSCKF with its seed"}
+ - {file: tests/tools/invariant_backend_adapter.cpp, case: "invariant adapter: the seed's tilt-bias correlation reaches the invariant prior"}
status: implemented
- note: "#608"
- deviations:
- - "the base's prior is itself an isotropic σ = 0.1 with no tilt-bias correlation (#609); this node maps it faithfully"
+ note: "#608; since #609 the base's prior correlates tilt and accelerometer bias, and J carries the correlation (δθ_world = R̂·δθ_body, b_a unchanged)"
- id: propagate
op: replace
@@ -270,7 +269,7 @@ system_invariants:
text: told the true noise, the filter improves and its yaw σ never collapses (EuRoC ideal tracks)
check: {file: scripts/euroc-consistency.sh}
status: failing
- note: "with one truth and S2 integrating both samples (#616, median NIS 1.00): fails 1 (2 of 66 at 1 px, MH_03 and MH_04, the same draws as the MSCKF: the shared tilt and bias seed, #609), passes 2 (0.40× vs 0.41×; 0.18× vs 0.36×), passes 3 (0.984)"
+ note: "with one truth, S2 integrating both samples (#616) and the tilt–bias seed prior (#609), median NIS 1.00: fails 1 (2 of 66 at 1 px, MH_03 and MH_04, the same draws as the MSCKF: moving starts' velocity seeds, not the tilt), passes 2 (0.32× vs 0.53×; 0.19× vs 0.40×), passes 3 (0.969; V2_01 and V2_02 fall to 0.64–0.67 at 0.5 px, #620)"
- id: accuracy
text: 4-DoF ATE and rotation RPE within a factor of OpenVINS mono on the 11 EuRoC sequences
check: {file: scripts/euroc-baseline.sh}
diff --git a/docs/pipeline/stages/s1-initialization.mdx b/docs/pipeline/stages/s1-initialization.mdx
index cedf9f6..0e51b9a 100644
--- a/docs/pipeline/stages/s1-initialization.mdx
+++ b/docs/pipeline/stages/s1-initialization.mdx
@@ -1,6 +1,6 @@
---
title: S1 — Initialization
-description: Bootstrapping attitude, biases, gravity, and metric scale — plus the initial covariance. The algorithm checks out; the isotropic covariance seed is the over-confidence suspect.
+description: Bootstrapping attitude, biases, gravity, and metric scale — plus the initial covariance, which since #609 correlates tilt with accelerometer bias as a static window implies.
---
import { Aside } from '@astrojs/starlight/components';
@@ -51,7 +51,7 @@ $$
* **Orthonormalization Singularity & Fallback**: If the gravity vector $\mathbf{z}_I$ is parallel or near-parallel to the temporary horizontal axis $\mathbf{x}_{\text{tmp}} = [1, 0, 0]^T$ (which happens if the drone is tilted exactly $90^\circ$ on its side), the cross product $\mathbf{z}_I \times \mathbf{x}_{\text{tmp}}$ collapses to zero, making step 2 undefined. Cortex solves this by checking the norm of $\mathbf{z}_I \times \mathbf{x}_{\text{tmp}}$. If it falls below a strict threshold (e.g., $0.1$), it dynamically falls back to using $\mathbf{x}_{\text{tmp}} = [0, 1, 0]^T$ as the temporary axis, ensuring the basis construction is always well-conditioned.
-Because gravity has zero horizontal components, **static accelerometer readings can only observe the two tilt axes (roll and pitch)**. The heading (**yaw**) is completely unobservable relative to gravity. Yaw is a free gauge direction; we can set it to an arbitrary starting value ($0^\circ$), but the filter must be told that this yaw direction is completely unobserved.
+Because gravity has zero horizontal components, **static accelerometer readings can only observe the two tilt axes (roll and pitch)**, and those only if the accelerometer bias is known: with an unknown horizontal bias the window sees tilt and bias as one sum (section 3). The heading (**yaw**) is completely unobservable relative to gravity. Yaw is a free gauge direction; we can set it to an arbitrary starting value ($0^\circ$), but the filter must be told that this yaw direction is completely unobserved.
---
@@ -79,28 +79,67 @@ To prevent the EKF from starting with a confidently wrong metric scale, Cortex r
---
-## 3. The Covariance Seed Suspect: Isotropic Over-Confidence
+## 3. The Covariance Seed: What the Window Determines
-The initial covariance $\mathbf{P}_0$ is a 15 x 15 matrix that represents the filter's starting uncertainty.
+The initial covariance $\mathbf{P}_0$ is the filter's starting uncertainty. Until #609 cortex
+seeded it isotropic, $\mathbf{P}_0 = \sigma^2 \mathbf{I}$ with $\sigma = 0.1$ on every core state, and
+that prior made a claim the window doesn't support: that tilt and accelerometer bias are
+independent.
-Historically, many VIO implementations (including those generated by AI Code Assistants) initialize $\mathbf{P}_0$ as an **isotropic matrix**:
+### Tilt and bias are one quantity at rest
+
+At rest the accelerometer reads $\mathbf{f} = \mathbf{R}^T g\,\mathbf{e}_z + \mathbf{b}_a + \mathbf{n}$.
+A seed that takes "up" $\hat{\mathbf{u}}$ from the mean $\bar{\mathbf{f}}$ with $\mathbf{b}_a = 0$
+leaves a tilt error ($\mathbf{R} = \hat{\mathbf{R}}\,\mathrm{Exp}(\delta\boldsymbol\theta)$) of
$$
-\mathbf{P}_0 = \sigma^2 \mathbf{I}
+\delta\boldsymbol\theta_\perp = [\hat{\mathbf{u}}]_\times (\mathbf{b}_a + \bar{\mathbf{n}}) / g ,
$$
-Where every block (roll, pitch, yaw, velocity, position, gyroscope bias, accelerometer bias) gets the same small initial uncertainty (e.g., $\sigma = 5.73^\circ$ for rotation).
-
-This isotropic seed is **mathematically dishonest and a major source of over-confidence**:
+so the window knows the tilt and the horizontal bias only as one sum. With $\mathbf{b}_a \sim
+\mathcal{N}(0, \sigma_{ba}^2 \mathbf{I})$ and the mean's own error $\sigma_f$ the seed's block is
-| State Block | Physical Observation at Static Init | Isotropic Seed ($\sigma = 5.73^\circ$) | Physically Honest Seed |
-| :--- | :--- | :--- | :--- |
-| **Roll / Pitch** | **Observed** via gravity-leveling ($\mathbf{a}_{\text{avg}}$). | $5.73^\circ$ | **Small** ($\approx 0.1^\circ$ to $0.5^\circ$) |
-| **Yaw (Heading)** | **Unobserved** (gravity has no horizontal component). | $5.73^\circ$ | **Huge** ($\approx 180^\circ$ or treated as infinite) |
-| **Velocity** | **Observed** (stationary table is $\mathbf{v} = \mathbf{0}$). | $0.1\text{ m/s}$ | **Small** ($\approx 10^{-4}\text{ m/s}$) |
-| **Accel Bias** | **Weakly Observed** over a static window. | $0.1\text{ m/s}^2$ | **Large** ($\approx 1.0\text{ m/s}^2$) |
+$$
+\mathbf{P}_{\theta\theta} = \tfrac{\sigma_{ba}^2 + \sigma_f^2}{g^2}(\mathbf{I} - \hat{\mathbf{u}}\hat{\mathbf{u}}^T) + \sigma_\psi^2 \hat{\mathbf{u}}\hat{\mathbf{u}}^T,\quad
+\mathbf{P}_{\theta b} = \tfrac{\sigma_{ba}^2}{g}[\hat{\mathbf{u}}]_\times,\quad
+\mathbf{P}_{bb} = \sigma_{ba}^2(\mathbf{I} - \hat{\mathbf{u}}\hat{\mathbf{u}}^T) + \sigma_\parallel^2 \hat{\mathbf{u}}\hat{\mathbf{u}}^T .
+$$
-By claiming that the unobservable yaw heading is known to $\pm5.73^\circ$ at $t = 0$, the isotropic seed injects false information into the EKF. Because the filter "thinks" its heading is highly accurate, it will reject visual measurements that deviate slightly due to camera rotation, leading directly to the **attitude-localized over-confidence** seen on the [Consistency Analysis](../consistency-analysis.mdx) page.
+The isotropic prior put 99% of the sum in tilt, because 0.1 rad of tilt is 1 m/s² of gravity
+against 0.1 m/s² of bias. Since #609 S1 writes this block on a static seed
+(`stages::s1_initialization::tilt_bias_prior`), with:
+
+| Term | Value | Why |
+| :--- | :--- | :--- |
+| $\sigma_{ba}$ | 0.2 m/s² (`VioConfig::accel_bias_prior_sigma`) | EuRoC's ground-truth bias is 0.09–0.20 m/s² (V1_01's 0.55 is suspect, below) |
+| $\sigma_f$ | the window's scatter over $\sqrt N$, or $\sigma_a/\sqrt{T}$ if larger | rotor vibration (0.03–0.28 m/s² per sample on EuRoC) |
+| bias along up | seeded from $\lvert\bar{\mathbf{f}}\rvert - g$, $\sigma_\parallel^2 = \sigma_f^2 + 0.026^2$ | with $g$ known the window measures it; 0.026 is half the spread of surface gravity |
+| $\sigma_\psi$ (yaw) | unchanged, 0.1 | a gauge: the gate watches that its σ never falls |
+
+The tilt σ that follows is $\sigma_{ba}/g \approx 1.2°$ to first order; the mean's own error
+$\sigma_f$ adds a few percent on a vibrating window. At rest, the accelerometer alone separates
+the two only through roll and pitch excursions; a turn about gravity leaves the body's up where it
+was. Once vision pins the position, motion separates them too: a body-fixed bias turns with the
+body while a tilt error stays fixed in the world, and an acceleration the camera sees is not
+gravity. The synthetic world's 0.5 m/s² bias is learned with yaw-only rotation and a moving loop;
+a platform that hovers level, without turning or accelerating, keeps the sum.
+
+### What it fixed, and what it didn't
+
+- **The seed is consistent:** over 600 Monte-Carlo static seeds the 5-dof tilt-and-bias ANEES
+ measured 4.93 against 5 (`tests/sdk/tilt_bias_prior.cpp` asserts 4.5–5.5). On the synthetic world a 0.5 m/s² bias is
+ learned and the tilt falls under 0.01°.
+- **V1_01's 2.4° gap to the ground truth stays.** The filter reaches the same bias (b_a y ≈ 0.13
+ m/s²) with $\sigma_{ba}$ = 0.2, 1.0 and the old prior, the value it finds on all eleven sequences.
+ The gap is a fixed rotation in the body frame, and the ground truth's 0.55 m/s² bias is its own
+ estimator's output: the data supports the filter.
+- **Moving starts keep the isotropic prior.** A gravity-aligned window accelerates, so its honest
+ tilt σ is about 15°; with that, the first updates explained the seed's wrong zero velocity by
+ tilt, and three of the gate's five MH_03 draws diverged against one. A dynamic seed's tilt comes
+ from the visual-inertial alignment (0.14° off on the gate's MH_04 draw); its error is the
+ seeded velocity.
+- **Open:** the gyro-bias prior is still the isotropic 0.1 rad/s although the static window
+ measures it, so the tilt σ grows by about 0.1 rad/s × the time to the first update.
---
@@ -110,9 +149,9 @@ The mathematical contracts ensuring safe filter bootstrapping:
| Element | Contract Specification |
| :--- | :--- |
-| **Signature** | $\text{init}(\text{imu\_window}, \text{vision\_window}) \rightarrow (\mathbf{R}_{I}^{W}, v, b_g, b_a, g, \text{scale})$ | $\mathbf{P}_0$ |
+| **Signature** | $\text{init}(\text{imu\_window}, \text{vision\_window}) \rightarrow (\mathbf{R}_{I}^{W}, v, b_g, b_a, g, \text{scale}, \mathbf{P}_0)$ |
| **Pre-conditions** | Gyro/accel variance is below stationary noise threshold (static init); singular-value conditioning of SfM-IMU matrix is below 100 (dynamic init). |
-| **Post-conditions & Invariants** | **1. Gravity alignment**: Gravity vector aligns with measured mean specific force: $\|\mathbf{R}_{I}^{W} \mathbf{a}_{\text{avg}} - [0, 0, g]^T\| < 10^{-4}\text{ m/s}^2$.
**2. Positive Scale**: Dynamically recovered scale $s$ is strictly positive ($s > 0$) and finite.
**3. Realistic Yaw Prior**: Heading (yaw) standard deviation is seeded with a large variance: $\text{std\_dev}(yaw) \ge 45^\circ$ to reflect its unobserved status. |
+| **Post-conditions & Invariants** | **1. Gravity alignment**: Gravity vector aligns with measured mean specific force: $\|\mathbf{R}_{I}^{W} \mathbf{a}_{\text{avg}} - [0, 0, g]^T\| < 10^{-4}\text{ m/s}^2$.
**2. Positive Scale**: Dynamically recovered scale $s$ is strictly positive ($s > 0$) and finite.
**3. Realistic Yaw Prior**: Heading (yaw) standard deviation is seeded with a large variance: $\text{std\_dev}(yaw) \ge 45^\circ$ to reflect its unobserved status (not met: the shipped yaw σ is 0.1 rad).
**4. Tilt–bias correlation**: a static seed's $\mathbf{P}_{\theta b} = \sigma_{ba}^2/g\,[\hat{\mathbf{u}}]_\times$, its 5-dof tilt-and-bias ANEES 5 (#609). |
---
diff --git a/scripts/euroc-baseline.json b/scripts/euroc-baseline.json
index efa10e0..98d419f 100644
--- a/scripts/euroc-baseline.json
+++ b/scripts/euroc-baseline.json
@@ -1,9 +1,9 @@
{
"_about": "EuRoC MAV baseline (docs/benchmarks/euroc-baseline.mdx). Aggregate metrics only, scored by tools/traj_eval: ATE in m (4-DoF and SE3 alignment), RPE as the mean over 8/16/24/32/40 m segments (m, deg), start_s and coverage_s in s after the first ground-truth stamp. scripts/euroc-baseline.sh reruns cortex and compares against cortex.run.",
"cortex": {
- "commit": "f1551cc",
+ "commit": "7a38162",
"binary": "tools/vio_pipeline --source euroc (MSCKF, double stages, full covariance)",
- "date": "2026-11-09"
+ "date": "2026-11-10"
},
"openvins": {
"release": "v2.7",
@@ -156,18 +156,18 @@
"compare_from_s": 19.153,
"cortex": {
"run": {
- "ate_4dof": 0.3753,
- "ate_se3": 0.3713,
- "rpe_trans_m": 0.6613,
- "rpe_rot_deg": 1.4864,
+ "ate_4dof": 0.3838,
+ "ate_se3": 0.3801,
+ "rpe_trans_m": 0.6199,
+ "rpe_rot_deg": 1.0075,
"start_s": 0.005,
"coverage_s": 98.75
},
"window": {
- "ate_4dof": 0.3785,
- "ate_se3": 0.3722,
+ "ate_4dof": 0.3848,
+ "ate_se3": 0.379,
"rpe_trans_m": 0.5477,
- "rpe_rot_deg": 1.0341,
+ "rpe_rot_deg": 0.7567,
"start_s": 19.155,
"coverage_s": 79.6
}
@@ -203,18 +203,18 @@
"compare_from_s": 19.734,
"cortex": {
"run": {
- "ate_4dof": 0.6714,
- "ate_se3": 0.6701,
- "rpe_trans_m": 0.8235,
- "rpe_rot_deg": 2.0584,
+ "ate_4dof": 0.686,
+ "ate_se3": 0.6847,
+ "rpe_trans_m": 0.8197,
+ "rpe_rot_deg": 1.8943,
"start_s": -0.015,
"coverage_s": 111.05
},
"window": {
- "ate_4dof": 0.6291,
- "ate_se3": 0.6268,
- "rpe_trans_m": 0.7012,
- "rpe_rot_deg": 0.9463,
+ "ate_4dof": 0.6396,
+ "ate_se3": 0.6376,
+ "rpe_trans_m": 0.722,
+ "rpe_rot_deg": 0.8678,
"start_s": 19.735,
"coverage_s": 91.3
}
@@ -250,18 +250,18 @@
"compare_from_s": 4.559,
"cortex": {
"run": {
- "ate_4dof": 0.2003,
- "ate_se3": 0.1987,
- "rpe_trans_m": 0.3509,
- "rpe_rot_deg": 4.0339,
+ "ate_4dof": 0.1922,
+ "ate_se3": 0.1907,
+ "rpe_trans_m": 0.3336,
+ "rpe_rot_deg": 3.9448,
"start_s": 0.01,
"coverage_s": 143.55
},
"window": {
- "ate_4dof": 0.1965,
- "ate_se3": 0.1952,
- "rpe_trans_m": 0.3296,
- "rpe_rot_deg": 4.0397,
+ "ate_4dof": 0.1881,
+ "ate_se3": 0.187,
+ "rpe_trans_m": 0.318,
+ "rpe_rot_deg": 4.0204,
"start_s": 4.56,
"coverage_s": 139.0
}
@@ -297,18 +297,18 @@
"compare_from_s": 3.955,
"cortex": {
"run": {
- "ate_4dof": 0.1625,
- "ate_se3": 0.1605,
- "rpe_trans_m": 0.2101,
- "rpe_rot_deg": 2.6787,
+ "ate_4dof": 0.1546,
+ "ate_se3": 0.1526,
+ "rpe_trans_m": 0.2141,
+ "rpe_rot_deg": 1.9737,
"start_s": 0.005,
"coverage_s": 83.5
},
"window": {
- "ate_4dof": 0.1623,
- "ate_se3": 0.1607,
- "rpe_trans_m": 0.2088,
- "rpe_rot_deg": 2.3128,
+ "ate_4dof": 0.1529,
+ "ate_se3": 0.1513,
+ "rpe_trans_m": 0.2079,
+ "rpe_rot_deg": 1.8734,
"start_s": 4.005,
"coverage_s": 79.5
}
@@ -344,18 +344,18 @@
"compare_from_s": 5.405,
"cortex": {
"run": {
- "ate_4dof": 0.1578,
- "ate_se3": 0.157,
- "rpe_trans_m": 0.2528,
- "rpe_rot_deg": 3.2681,
+ "ate_4dof": 0.1447,
+ "ate_se3": 0.1438,
+ "rpe_trans_m": 0.2296,
+ "rpe_rot_deg": 2.7062,
"start_s": 0.005,
"coverage_s": 104.65
},
"window": {
- "ate_4dof": 0.1612,
- "ate_se3": 0.1601,
- "rpe_trans_m": 0.2351,
- "rpe_rot_deg": 2.5587,
+ "ate_4dof": 0.147,
+ "ate_se3": 0.1456,
+ "rpe_trans_m": 0.207,
+ "rpe_rot_deg": 1.9323,
"start_s": 5.405,
"coverage_s": 99.25
}
@@ -391,18 +391,18 @@
"compare_from_s": 3.725,
"cortex": {
"run": {
- "ate_4dof": 0.2233,
- "ate_se3": 0.222,
- "rpe_trans_m": 0.4494,
- "rpe_rot_deg": 2.8327,
+ "ate_4dof": 0.191,
+ "ate_se3": 0.1898,
+ "rpe_trans_m": 0.374,
+ "rpe_rot_deg": 1.9732,
"start_s": 0.025,
"coverage_s": 111.95
},
"window": {
- "ate_4dof": 0.2148,
- "ate_se3": 0.2134,
- "rpe_trans_m": 0.4472,
- "rpe_rot_deg": 2.4058,
+ "ate_4dof": 0.184,
+ "ate_se3": 0.1827,
+ "rpe_trans_m": 0.3705,
+ "rpe_rot_deg": 1.942,
"start_s": 3.725,
"coverage_s": 108.25
}
@@ -438,18 +438,18 @@
"compare_from_s": 2.88,
"cortex": {
"run": {
- "ate_4dof": 0.3381,
- "ate_se3": 0.331,
- "rpe_trans_m": 0.3063,
- "rpe_rot_deg": 4.3461,
+ "ate_4dof": 0.327,
+ "ate_se3": 0.3198,
+ "rpe_trans_m": 0.2746,
+ "rpe_rot_deg": 3.0305,
"start_s": -0.02,
"coverage_s": 115.45
},
"window": {
- "ate_4dof": 0.3335,
- "ate_se3": 0.3258,
- "rpe_trans_m": 0.3011,
- "rpe_rot_deg": 4.0922,
+ "ate_4dof": 0.3221,
+ "ate_se3": 0.3144,
+ "rpe_trans_m": 0.2724,
+ "rpe_rot_deg": 3.0271,
"start_s": 2.93,
"coverage_s": 112.5
}
@@ -485,18 +485,18 @@
"compare_from_s": 4.565,
"cortex": {
"run": {
- "ate_4dof": 0.1728,
- "ate_se3": 0.1686,
- "rpe_trans_m": 0.2247,
- "rpe_rot_deg": 1.957,
+ "ate_4dof": 0.1743,
+ "ate_se3": 0.1698,
+ "rpe_trans_m": 0.2186,
+ "rpe_rot_deg": 1.4725,
"start_s": 0.415,
"coverage_s": 114.4
},
"window": {
- "ate_4dof": 0.175,
- "ate_se3": 0.1711,
- "rpe_trans_m": 0.225,
- "rpe_rot_deg": 1.7371,
+ "ate_4dof": 0.177,
+ "ate_se3": 0.1727,
+ "rpe_trans_m": 0.2222,
+ "rpe_rot_deg": 1.4959,
"start_s": 4.565,
"coverage_s": 110.25
}
diff --git a/scripts/euroc-consistency.json b/scripts/euroc-consistency.json
index 273d610..765107e 100644
--- a/scripts/euroc-consistency.json
+++ b/scripts/euroc-consistency.json
@@ -6,25 +6,25 @@
},
"backends": {
"msckf": {
- "commit": "f1551cc",
- "date": "2026-11-09",
+ "commit": "7a38162",
+ "date": "2026-11-10",
"verdict": {
"no_divergence": false,
"matched_beats_default": true,
"no_yaw_collapse": false,
"pass": false,
- "median_yaw_sigma_ratio": 0.1787,
- "median_yaw_nees": 7.674,
+ "median_yaw_sigma_ratio": 0.1521,
+ "median_yaw_nees": 8.244,
"median_nis": 1.002,
"truth": "spline (--imu gt, #613)",
"median_ratio": {
"1": {
- "matched": 0.452,
- "default": 0.519
+ "matched": 0.36,
+ "default": 0.638
},
"0.5": {
- "matched": 0.327,
- "default": 0.529
+ "matched": 0.249,
+ "default": 0.577
}
},
"diverged": [
@@ -803,1370 +803,1370 @@
},
"MH_05_difficult/1/1/matched": {
"status": "scored",
- "ate_4dof": 0.0401,
- "ratio": 0.088,
- "rpe_rot_deg": 0.165,
- "nees_yaw": 117.12028518520545,
- "nees_rp": 3.8633023577972896,
- "nees_pos": 27.604300437728405,
- "yaw_sigma_first": 0.1000799729890207,
- "yaw_sigma_min": 0.006451956718817454,
- "yaw_sigma_ratio": 0.0645,
- "yaw_err_max": 0.0860234284524179,
- "nis": 1.00592,
- "ledger_info_yaw": 3353349766.9891887,
- "ledger_share_s6": 3353349766.980577
+ "ate_4dof": 0.0423,
+ "ratio": 0.092,
+ "rpe_rot_deg": 0.307,
+ "nees_yaw": 14.721923452679228,
+ "nees_rp": 2.101410347154978,
+ "nees_pos": 10.278140881531996,
+ "yaw_sigma_first": 0.10007997298902069,
+ "yaw_sigma_min": 0.020587696621608424,
+ "yaw_sigma_ratio": 0.2057,
+ "yaw_err_max": 0.1452209725573232,
+ "nis": 1.00239,
+ "ledger_info_yaw": 537307748.4024909,
+ "ledger_share_s6": 537307748.405568
},
"MH_05_difficult/1/1/default": {
"status": "scored",
- "ate_4dof": 0.0494,
- "ratio": 0.108,
- "rpe_rot_deg": 0.304,
- "nees_yaw": 0.05077973651042521,
- "nees_rp": 1.3842741099379425,
- "nees_pos": 1.0014497876491804,
- "yaw_sigma_first": 0.1000799729890207,
- "yaw_sigma_min": 0.05910955597396117,
- "yaw_sigma_ratio": 0.5906,
- "yaw_err_max": 0.023320925639997703,
- "nis": 0.0478021,
- "ledger_info_yaw": 2697101.9474149677,
- "ledger_share_s6": 2697101.947372601
+ "ate_4dof": 0.0503,
+ "ratio": 0.11,
+ "rpe_rot_deg": 0.25,
+ "nees_yaw": 0.3273629175903739,
+ "nees_rp": 1.6019512905468813,
+ "nees_pos": 0.8242316305057027,
+ "yaw_sigma_first": 0.10007997298902069,
+ "yaw_sigma_min": 0.03431282267571579,
+ "yaw_sigma_ratio": 0.3429,
+ "yaw_err_max": 0.035570801154258744,
+ "nis": 0.0480261,
+ "ledger_info_yaw": 39640913.59069673,
+ "ledger_share_s6": 39640913.59091514
},
"MH_05_difficult/1/2/matched": {
"status": "scored",
- "ate_4dof": 0.0467,
- "ratio": 0.102,
- "rpe_rot_deg": 0.16,
- "nees_yaw": 65.26143526300241,
- "nees_rp": 2.111474236232537,
- "nees_pos": 36.6374408175422,
- "yaw_sigma_first": 0.10007997298902244,
- "yaw_sigma_min": 0.013384953452065076,
- "yaw_sigma_ratio": 0.1337,
- "yaw_err_max": 0.13969661883473825,
- "nis": 1.00661,
- "ledger_info_yaw": 317190244.91006404,
- "ledger_share_s6": 317190244.9100687
+ "ate_4dof": 0.0464,
+ "ratio": 0.101,
+ "rpe_rot_deg": 0.18,
+ "nees_yaw": 13.633453203830307,
+ "nees_rp": 1.4230891685142792,
+ "nees_pos": 6.080517456623719,
+ "yaw_sigma_first": 0.1000799729890225,
+ "yaw_sigma_min": 0.015925649015123736,
+ "yaw_sigma_ratio": 0.1591,
+ "yaw_err_max": 0.07117343253650917,
+ "nis": 1.00171,
+ "ledger_info_yaw": 70641123.53881826,
+ "ledger_share_s6": 70641123.53910583
},
"MH_05_difficult/1/2/default": {
"status": "scored",
- "ate_4dof": 0.0976,
- "ratio": 0.213,
- "rpe_rot_deg": 0.309,
- "nees_yaw": 144.34828988149079,
- "nees_rp": 2.654397645354802,
- "nees_pos": 75.74572492978531,
- "yaw_sigma_first": 0.10007997298902244,
- "yaw_sigma_min": 0.015838629274815427,
- "yaw_sigma_ratio": 0.1583,
- "yaw_err_max": 0.275059810433913,
- "nis": 0.0494715,
- "ledger_info_yaw": 25684741.84131491,
- "ledger_share_s6": 25684741.841334574
+ "ate_4dof": 0.0851,
+ "ratio": 0.186,
+ "rpe_rot_deg": 0.342,
+ "nees_yaw": 0.17551235642477783,
+ "nees_rp": 1.0019834883501433,
+ "nees_pos": 0.4030929445550768,
+ "yaw_sigma_first": 0.1000799729890225,
+ "yaw_sigma_min": 0.022460942146854405,
+ "yaw_sigma_ratio": 0.2244,
+ "yaw_err_max": 0.056971414005004245,
+ "nis": 0.0479581,
+ "ledger_info_yaw": 411228312.6596239,
+ "ledger_share_s6": 411228312.6601435
},
"MH_05_difficult/1/3/matched": {
"status": "scored",
- "ate_4dof": 0.0442,
- "ratio": 0.097,
- "rpe_rot_deg": 0.231,
- "nees_yaw": 92.1113227981576,
- "nees_rp": 24.409251368988144,
- "nees_pos": 17.484357522445713,
- "yaw_sigma_first": 0.10007997298900424,
- "yaw_sigma_min": 0.006990297981878217,
- "yaw_sigma_ratio": 0.0698,
- "yaw_err_max": 0.09616825131525158,
- "nis": 1.01736,
- "ledger_info_yaw": 31739503433.7255,
- "ledger_share_s6": 31739503433.75392
+ "ate_4dof": 0.0423,
+ "ratio": 0.093,
+ "rpe_rot_deg": 0.299,
+ "nees_yaw": 2.827353782414646,
+ "nees_rp": 2.210616604625209,
+ "nees_pos": 0.9600533187093141,
+ "yaw_sigma_first": 0.10007997298900428,
+ "yaw_sigma_min": 0.01931518336572964,
+ "yaw_sigma_ratio": 0.193,
+ "yaw_err_max": 0.11442185368861531,
+ "nis": 1.00182,
+ "ledger_info_yaw": 240881083.6678422,
+ "ledger_share_s6": 240881083.66882908
},
"MH_05_difficult/1/3/default": {
"status": "scored",
- "ate_4dof": 0.0658,
- "ratio": 0.144,
- "rpe_rot_deg": 0.379,
- "nees_yaw": 0.5144396421739174,
- "nees_rp": 1.8207994207052216,
- "nees_pos": 0.6400289140520988,
- "yaw_sigma_first": 0.10007997298900424,
- "yaw_sigma_min": 0.04529475755421575,
- "yaw_sigma_ratio": 0.4526,
- "yaw_err_max": 0.043836672234813745,
- "nis": 0.0478494,
- "ledger_info_yaw": 36347443.22012914,
- "ledger_share_s6": 36347443.22001146
+ "ate_4dof": 0.0639,
+ "ratio": 0.14,
+ "rpe_rot_deg": 0.418,
+ "nees_yaw": 5.770644299135114,
+ "nees_rp": 2.1058477507757862,
+ "nees_pos": 5.628394118892647,
+ "yaw_sigma_first": 0.10007997298900428,
+ "yaw_sigma_min": 0.031556900825354055,
+ "yaw_sigma_ratio": 0.3153,
+ "yaw_err_max": 0.1070828218405009,
+ "nis": 0.0479576,
+ "ledger_info_yaw": 145282575.3134606,
+ "ledger_share_s6": 145282575.31380653
},
"MH_05_difficult/0.5/1/matched": {
"status": "scored",
- "ate_4dof": 0.0367,
- "ratio": 0.08,
- "rpe_rot_deg": 0.325,
- "nees_yaw": 22.796853969325067,
- "nees_rp": 2.5992207711253146,
- "nees_pos": 13.613930822807108,
- "yaw_sigma_first": 0.1000799729890207,
- "yaw_sigma_min": 0.020680471997558212,
- "yaw_sigma_ratio": 0.2066,
- "yaw_err_max": 0.14943146023420587,
- "nis": 1.00215,
- "ledger_info_yaw": 1458464306.0310035,
- "ledger_share_s6": 1565458444.0095046
+ "ate_4dof": 0.0259,
+ "ratio": 0.057,
+ "rpe_rot_deg": 0.224,
+ "nees_yaw": 8.57128152591674,
+ "nees_rp": 3.0138368149747965,
+ "nees_pos": 4.410813276416057,
+ "yaw_sigma_first": 0.10007997298902069,
+ "yaw_sigma_min": 0.017432136618492168,
+ "yaw_sigma_ratio": 0.1742,
+ "yaw_err_max": 0.09241001516752217,
+ "nis": 1.00217,
+ "ledger_info_yaw": 1376449111.5494409,
+ "ledger_share_s6": 1346963110.9619672
},
"MH_05_difficult/0.5/1/default": {
"status": "scored",
- "ate_4dof": 0.0692,
- "ratio": 0.151,
- "rpe_rot_deg": 0.451,
- "nees_yaw": 2.676832507018515,
- "nees_rp": 1.9074368593117732,
- "nees_pos": 1.5771118026137319,
- "yaw_sigma_first": 0.1000799729890207,
- "yaw_sigma_min": 0.030545662054482982,
- "yaw_sigma_ratio": 0.3052,
- "yaw_err_max": 0.07586199753788722,
- "nis": 0.0121326,
- "ledger_info_yaw": 56388140.558103226,
- "ledger_share_s6": 48480378.07263467
+ "ate_4dof": 0.0644,
+ "ratio": 0.141,
+ "rpe_rot_deg": 0.41,
+ "nees_yaw": 4.54450061919466,
+ "nees_rp": 2.008813704123303,
+ "nees_pos": 1.9617446926028677,
+ "yaw_sigma_first": 0.10007997298902069,
+ "yaw_sigma_min": 0.017343457553562314,
+ "yaw_sigma_ratio": 0.1733,
+ "yaw_err_max": 0.07864476115915238,
+ "nis": 0.012139,
+ "ledger_info_yaw": 94933942.1916304,
+ "ledger_share_s6": 79754224.89416234
},
"MH_05_difficult/0.5/2/matched": {
"status": "scored",
- "ate_4dof": 0.0458,
- "ratio": 0.1,
- "rpe_rot_deg": 0.342,
- "nees_yaw": 4.661679759175424,
- "nees_rp": 1.4337419879151598,
- "nees_pos": 2.8879305386981016,
- "yaw_sigma_first": 0.10007997298902244,
- "yaw_sigma_min": 0.023060410053287853,
- "yaw_sigma_ratio": 0.2304,
- "yaw_err_max": 0.0815894713160032,
- "nis": 1.00098,
- "ledger_info_yaw": 758996264.1697748,
- "ledger_share_s6": 950619335.4927443
+ "ate_4dof": 0.0342,
+ "ratio": 0.075,
+ "rpe_rot_deg": 0.22,
+ "nees_yaw": 3.8298907227930736,
+ "nees_rp": 1.4976532965947515,
+ "nees_pos": 2.045206907580843,
+ "yaw_sigma_first": 0.1000799729890225,
+ "yaw_sigma_min": 0.019147533124505827,
+ "yaw_sigma_ratio": 0.1913,
+ "yaw_err_max": 0.08453850344622982,
+ "nis": 1.00104,
+ "ledger_info_yaw": 2047660382.750858,
+ "ledger_share_s6": 1789761777.6698964
},
"MH_05_difficult/0.5/2/default": {
"status": "scored",
- "ate_4dof": 0.1099,
- "ratio": 0.24,
- "rpe_rot_deg": 0.38,
- "nees_yaw": 2.632603094867554,
- "nees_rp": 0.6709087294057351,
- "nees_pos": 1.993729572960375,
- "yaw_sigma_first": 0.10007997298902244,
- "yaw_sigma_min": 0.04987415832449594,
- "yaw_sigma_ratio": 0.4983,
- "yaw_err_max": 0.09720453196419604,
- "nis": 0.0120759,
- "ledger_info_yaw": 16060235.235171977,
- "ledger_share_s6": 17093505.347556736
+ "ate_4dof": 0.1055,
+ "ratio": 0.231,
+ "rpe_rot_deg": 0.372,
+ "nees_yaw": 6.458399853055387,
+ "nees_rp": 0.8932930230681245,
+ "nees_pos": 3.6118876517762213,
+ "yaw_sigma_first": 0.1000799729890225,
+ "yaw_sigma_min": 0.029465026522051085,
+ "yaw_sigma_ratio": 0.2944,
+ "yaw_err_max": 0.1483337536023303,
+ "nis": 0.0121044,
+ "ledger_info_yaw": 87149631.16976173,
+ "ledger_share_s6": 60738642.99391055
},
"MH_05_difficult/0.5/3/matched": {
"status": "scored",
- "ate_4dof": 0.043,
- "ratio": 0.094,
- "rpe_rot_deg": 0.297,
- "nees_yaw": 2.9625600800821266,
- "nees_rp": 1.8035395806663654,
- "nees_pos": 1.730306222838167,
- "yaw_sigma_first": 0.10007997298900424,
- "yaw_sigma_min": 0.018256561054389716,
- "yaw_sigma_ratio": 0.1824,
- "yaw_err_max": 0.04308610767210108,
- "nis": 1.00185,
- "ledger_info_yaw": 130983253.70659864,
- "ledger_share_s6": 180242993.36254328
+ "ate_4dof": 0.071,
+ "ratio": 0.155,
+ "rpe_rot_deg": 0.516,
+ "nees_yaw": 0.6349814163639929,
+ "nees_rp": 1.822457376971377,
+ "nees_pos": 0.4761694306667355,
+ "yaw_sigma_first": 0.10007997298900428,
+ "yaw_sigma_min": 0.022538544276496007,
+ "yaw_sigma_ratio": 0.2252,
+ "yaw_err_max": 0.05098493805981168,
+ "nis": 1.002,
+ "ledger_info_yaw": 137965462.8112977,
+ "ledger_share_s6": 156890764.89504227
},
"MH_05_difficult/0.5/3/default": {
"status": "scored",
- "ate_4dof": 0.0775,
- "ratio": 0.169,
- "rpe_rot_deg": 0.483,
- "nees_yaw": 2.4048374386264224,
- "nees_rp": 1.870597138516804,
- "nees_pos": 1.8137926024304947,
- "yaw_sigma_first": 0.10007997298900424,
- "yaw_sigma_min": 0.04932455241932518,
- "yaw_sigma_ratio": 0.4929,
- "yaw_err_max": 0.09414475657445229,
- "nis": 0.0121115,
- "ledger_info_yaw": 4449512.679771922,
- "ledger_share_s6": 3644840.360542276
+ "ate_4dof": 0.0666,
+ "ratio": 0.146,
+ "rpe_rot_deg": 0.348,
+ "nees_yaw": 6.8522300801993845,
+ "nees_rp": 2.287501827329675,
+ "nees_pos": 4.37686576791757,
+ "yaw_sigma_first": 0.10007997298900428,
+ "yaw_sigma_min": 0.03262183518125182,
+ "yaw_sigma_ratio": 0.326,
+ "yaw_err_max": 0.157024230158117,
+ "nis": 0.0121354,
+ "ledger_info_yaw": 26523654.24778022,
+ "ledger_share_s6": 16243167.95782458
},
"V1_01_easy/1/1/matched": {
"status": "scored",
- "ate_4dof": 0.0584,
- "ratio": 0.845,
- "rpe_rot_deg": 0.246,
- "nees_yaw": 9.272225138440279,
- "nees_rp": 1.8262199589297323,
- "nees_pos": 1.3697940292879507,
- "yaw_sigma_first": 0.10001125144439335,
- "yaw_sigma_min": 0.008159069467718962,
- "yaw_sigma_ratio": 0.0816,
- "yaw_err_max": 0.0759995468359201,
- "nis": 1.00548,
- "ledger_info_yaw": 2313090894.9604926,
- "ledger_share_s6": 2313090894.957644
+ "ate_4dof": 0.0315,
+ "ratio": 0.455,
+ "rpe_rot_deg": 0.334,
+ "nees_yaw": 93.45791146747642,
+ "nees_rp": 2.714135833648373,
+ "nees_pos": 14.34781644329224,
+ "yaw_sigma_first": 0.10001125144439336,
+ "yaw_sigma_min": 0.015874060155243026,
+ "yaw_sigma_ratio": 0.1587,
+ "yaw_err_max": 0.20925497115095346,
+ "nis": 0.999878,
+ "ledger_info_yaw": 13726117536.648203,
+ "ledger_share_s6": 13726117536.599232
},
"V1_01_easy/1/1/default": {
"status": "scored",
- "ate_4dof": 0.0517,
- "ratio": 0.748,
- "rpe_rot_deg": 0.291,
- "nees_yaw": 0.20800520244700837,
- "nees_rp": 1.718509355337741,
- "nees_pos": 0.3844748996013534,
- "yaw_sigma_first": 0.10001125144439335,
- "yaw_sigma_min": 0.039286915728142864,
- "yaw_sigma_ratio": 0.3928,
- "yaw_err_max": 0.038825692967054785,
- "nis": 0.0476938,
- "ledger_info_yaw": 50866515.98989653,
- "ledger_share_s6": 50866515.99006992
+ "ate_4dof": 0.0513,
+ "ratio": 0.743,
+ "rpe_rot_deg": 0.766,
+ "nees_yaw": 19.690236325429602,
+ "nees_rp": 3.647891361327173,
+ "nees_pos": 4.729956861691393,
+ "yaw_sigma_first": 0.10001125144439336,
+ "yaw_sigma_min": 0.024621537004683888,
+ "yaw_sigma_ratio": 0.2462,
+ "yaw_err_max": 0.19556182334518943,
+ "nis": 0.0479312,
+ "ledger_info_yaw": 299900929.6126127,
+ "ledger_share_s6": 299900929.61189485
},
"V1_01_easy/1/2/matched": {
"status": "scored",
- "ate_4dof": 0.1486,
- "ratio": 2.15,
- "rpe_rot_deg": 0.243,
- "nees_yaw": 10.525374329774877,
- "nees_rp": 2.083195850788314,
- "nees_pos": 0.8581160070501379,
+ "ate_4dof": 0.0279,
+ "ratio": 0.403,
+ "rpe_rot_deg": 0.183,
+ "nees_yaw": 25.734617264473968,
+ "nees_rp": 1.0482794684253478,
+ "nees_pos": 2.3767665010939067,
"yaw_sigma_first": 0.10001125144439178,
- "yaw_sigma_min": 0.006053904780425456,
- "yaw_sigma_ratio": 0.0605,
- "yaw_err_max": 0.03854400318411572,
- "nis": 1.00812,
- "ledger_info_yaw": 15828026801.158604,
- "ledger_share_s6": 15828026801.232172
+ "yaw_sigma_min": 0.013704978812945549,
+ "yaw_sigma_ratio": 0.137,
+ "yaw_err_max": 0.09428043931894972,
+ "nis": 1.00154,
+ "ledger_info_yaw": 33720106953.573227,
+ "ledger_share_s6": 33720106953.63016
},
"V1_01_easy/1/2/default": {
"status": "scored",
- "ate_4dof": 0.0393,
- "ratio": 0.568,
- "rpe_rot_deg": 0.532,
- "nees_yaw": 0.09508542285192181,
- "nees_rp": 0.7456698587739475,
- "nees_pos": 0.16636211820852817,
+ "ate_4dof": 0.0645,
+ "ratio": 0.933,
+ "rpe_rot_deg": 1.254,
+ "nees_yaw": 6.214595895225918,
+ "nees_rp": 3.3102870580670967,
+ "nees_pos": 2.1201757761004,
"yaw_sigma_first": 0.10001125144439178,
- "yaw_sigma_min": 0.05089791402653358,
- "yaw_sigma_ratio": 0.5089,
- "yaw_err_max": 0.040114274679714235,
- "nis": 0.0477479,
- "ledger_info_yaw": 266823135.38083667,
- "ledger_share_s6": 266823135.38067302
+ "yaw_sigma_min": 0.027406459995099525,
+ "yaw_sigma_ratio": 0.274,
+ "yaw_err_max": 0.195180007999386,
+ "nis": 0.047914,
+ "ledger_info_yaw": 4404067217.49089,
+ "ledger_share_s6": 4404067217.4866495
},
"V1_01_easy/1/3/matched": {
"status": "scored",
- "ate_4dof": 0.0285,
- "ratio": 0.413,
- "rpe_rot_deg": 0.539,
- "nees_yaw": 0.41286488018501505,
- "nees_rp": 2.290009655059887,
- "nees_pos": 0.14223225804347314,
- "yaw_sigma_first": 0.10001125144439336,
- "yaw_sigma_min": 0.018346942830047713,
- "yaw_sigma_ratio": 0.1834,
- "yaw_err_max": 0.036351338238245996,
- "nis": 1.00574,
- "ledger_info_yaw": 4947416598.23991,
- "ledger_share_s6": 4947416598.24957
+ "ate_4dof": 0.0289,
+ "ratio": 0.419,
+ "rpe_rot_deg": 0.622,
+ "nees_yaw": 8.211409585293513,
+ "nees_rp": 2.7581764738557455,
+ "nees_pos": 0.9436572410511962,
+ "yaw_sigma_first": 0.10001125144439339,
+ "yaw_sigma_min": 0.014067641911120463,
+ "yaw_sigma_ratio": 0.1407,
+ "yaw_err_max": 0.05426506196717375,
+ "nis": 1.00414,
+ "ledger_info_yaw": 4694981157.7484865,
+ "ledger_share_s6": 4694981157.722689
},
"V1_01_easy/1/3/default": {
"status": "scored",
- "ate_4dof": 0.0432,
- "ratio": 0.626,
- "rpe_rot_deg": 0.841,
- "nees_yaw": 1.4824832805335417,
- "nees_rp": 1.6588597591832757,
- "nees_pos": 0.7332281398064245,
- "yaw_sigma_first": 0.10001125144439336,
- "yaw_sigma_min": 0.042670941886363306,
- "yaw_sigma_ratio": 0.4267,
- "yaw_err_max": 0.07512736371626576,
- "nis": 0.0478586,
- "ledger_info_yaw": 5860460843.897238,
- "ledger_share_s6": 5860460843.890442
+ "ate_4dof": 0.0695,
+ "ratio": 1.005,
+ "rpe_rot_deg": 0.801,
+ "nees_yaw": 53.73990687279826,
+ "nees_rp": 3.7785435256386233,
+ "nees_pos": 11.099922377124619,
+ "yaw_sigma_first": 0.10001125144439339,
+ "yaw_sigma_min": 0.022810089081643007,
+ "yaw_sigma_ratio": 0.2281,
+ "yaw_err_max": 0.23659415121041613,
+ "nis": 0.0484557,
+ "ledger_info_yaw": 2627490708.0671525,
+ "ledger_share_s6": 2627490708.0648036
},
"V1_01_easy/0.5/1/matched": {
"status": "scored",
- "ate_4dof": 0.0237,
- "ratio": 0.343,
- "rpe_rot_deg": 0.379,
- "nees_yaw": 1.1897636523861337,
- "nees_rp": 1.5735222232002293,
- "nees_pos": 0.47552818257452123,
- "yaw_sigma_first": 0.10001125144439335,
- "yaw_sigma_min": 0.02104504458097203,
- "yaw_sigma_ratio": 0.2104,
- "yaw_err_max": 0.07171226410698069,
- "nis": 0.99978,
- "ledger_info_yaw": 175370259392.19623,
- "ledger_share_s6": 112055938409.26936
+ "ate_4dof": 0.0227,
+ "ratio": 0.329,
+ "rpe_rot_deg": 0.282,
+ "nees_yaw": 1.5044644360351158,
+ "nees_rp": 1.8431319555857333,
+ "nees_pos": 0.3465165259990266,
+ "yaw_sigma_first": 0.10001125144439336,
+ "yaw_sigma_min": 0.01617326241575341,
+ "yaw_sigma_ratio": 0.1617,
+ "yaw_err_max": 0.14807807971817316,
+ "nis": 0.999875,
+ "ledger_info_yaw": 266759022605.0859,
+ "ledger_share_s6": 110012523968.49686
},
"V1_01_easy/0.5/1/default": {
"status": "scored",
- "ate_4dof": 0.0443,
- "ratio": 0.64,
- "rpe_rot_deg": 1.144,
- "nees_yaw": 0.16461706286274777,
- "nees_rp": 1.8389353138368645,
- "nees_pos": 0.10471999274353748,
- "yaw_sigma_first": 0.10001125144439335,
- "yaw_sigma_min": 0.05900938084429211,
- "yaw_sigma_ratio": 0.59,
- "yaw_err_max": 0.05465711393079179,
- "nis": 0.0120785,
- "ledger_info_yaw": 137275647.22065157,
- "ledger_share_s6": 74592945.725698
+ "ate_4dof": 0.0421,
+ "ratio": 0.609,
+ "rpe_rot_deg": 0.647,
+ "nees_yaw": 3.391893727651269,
+ "nees_rp": 2.5606471151124026,
+ "nees_pos": 2.0505432522040183,
+ "yaw_sigma_first": 0.10001125144439336,
+ "yaw_sigma_min": 0.04332849377519899,
+ "yaw_sigma_ratio": 0.4332,
+ "yaw_err_max": 0.14754699854325531,
+ "nis": 0.0121073,
+ "ledger_info_yaw": 282798112.53465205,
+ "ledger_share_s6": 168923722.25764358
},
"V1_01_easy/0.5/2/matched": {
"status": "scored",
- "ate_4dof": 0.0327,
- "ratio": 0.473,
- "rpe_rot_deg": 0.906,
- "nees_yaw": 0.9033658749355722,
- "nees_rp": 1.0507125035963518,
- "nees_pos": 0.498349437919693,
+ "ate_4dof": 0.0248,
+ "ratio": 0.359,
+ "rpe_rot_deg": 0.43,
+ "nees_yaw": 4.112749584344797,
+ "nees_rp": 1.1521776014576433,
+ "nees_pos": 0.6781011217344377,
"yaw_sigma_first": 0.10001125144439178,
- "yaw_sigma_min": 0.025008944317136853,
- "yaw_sigma_ratio": 0.2501,
- "yaw_err_max": 0.05122106770410767,
- "nis": 1.00174,
- "ledger_info_yaw": 7499837178.044607,
- "ledger_share_s6": 2557444163.455643
+ "yaw_sigma_min": 0.018609765924571132,
+ "yaw_sigma_ratio": 0.1861,
+ "yaw_err_max": 0.09188306143183811,
+ "nis": 1.0006,
+ "ledger_info_yaw": 14070432959.412802,
+ "ledger_share_s6": 8586601113.594164
},
"V1_01_easy/0.5/2/default": {
"status": "scored",
- "ate_4dof": 0.0435,
- "ratio": 0.63,
- "rpe_rot_deg": 0.577,
- "nees_yaw": 0.5811008626300651,
- "nees_rp": 0.8036323981418818,
- "nees_pos": 0.6261657657689309,
+ "ate_4dof": 0.0414,
+ "ratio": 0.599,
+ "rpe_rot_deg": 0.452,
+ "nees_yaw": 1.8478679590885572,
+ "nees_rp": 1.3467472215662784,
+ "nees_pos": 0.7279051443182369,
"yaw_sigma_first": 0.10001125144439178,
- "yaw_sigma_min": 0.05822179128328464,
- "yaw_sigma_ratio": 0.5822,
- "yaw_err_max": 0.05888617733753346,
- "nis": 0.012099,
- "ledger_info_yaw": 199628237.79368445,
- "ledger_share_s6": 109422666.29361081
+ "yaw_sigma_min": 0.040181914813503986,
+ "yaw_sigma_ratio": 0.4018,
+ "yaw_err_max": 0.12578252303618842,
+ "nis": 0.012152,
+ "ledger_info_yaw": 222152349.352437,
+ "ledger_share_s6": 214443937.01095116
},
"V1_01_easy/0.5/3/matched": {
"status": "scored",
- "ate_4dof": 0.0292,
- "ratio": 0.423,
- "rpe_rot_deg": 1.074,
- "nees_yaw": 45.67815333488261,
- "nees_rp": 2.3428782744371177,
- "nees_pos": 12.072775576236431,
- "yaw_sigma_first": 0.10001125144439336,
- "yaw_sigma_min": 0.022637298604880337,
- "yaw_sigma_ratio": 0.2263,
- "yaw_err_max": 0.21222361076547203,
- "nis": 1.00363,
- "ledger_info_yaw": 1831671641.3239036,
- "ledger_share_s6": -798232647.1213629
+ "ate_4dof": 0.0188,
+ "ratio": 0.273,
+ "rpe_rot_deg": 0.552,
+ "nees_yaw": 90.88389006654052,
+ "nees_rp": 2.335754954790918,
+ "nees_pos": 13.667485655853168,
+ "yaw_sigma_first": 0.10001125144439339,
+ "yaw_sigma_min": 0.0162296140726989,
+ "yaw_sigma_ratio": 0.1623,
+ "yaw_err_max": 0.18357236545403885,
+ "nis": 1.00381,
+ "ledger_info_yaw": 11953430409.1656,
+ "ledger_share_s6": 7430065658.105801
},
"V1_01_easy/0.5/3/default": {
"status": "scored",
- "ate_4dof": 0.0446,
- "ratio": 0.645,
- "rpe_rot_deg": 1.057,
- "nees_yaw": 0.06163750654301657,
- "nees_rp": 1.8793087628519103,
- "nees_pos": 0.16530958528978676,
- "yaw_sigma_first": 0.10001125144439336,
- "yaw_sigma_min": 0.060977649233171456,
- "yaw_sigma_ratio": 0.6097,
- "yaw_err_max": 0.05303010855118145,
- "nis": 0.0121199,
- "ledger_info_yaw": 235936233.92949083,
- "ledger_share_s6": 135283893.19769216
+ "ate_4dof": 0.0414,
+ "ratio": 0.6,
+ "rpe_rot_deg": 0.499,
+ "nees_yaw": 3.254998833032602,
+ "nees_rp": 2.2055569845050282,
+ "nees_pos": 2.058516948885088,
+ "yaw_sigma_first": 0.10001125144439339,
+ "yaw_sigma_min": 0.04592327531081349,
+ "yaw_sigma_ratio": 0.4592,
+ "yaw_err_max": 0.14958086968385173,
+ "nis": 0.0121669,
+ "ledger_info_yaw": 80540562.75259411,
+ "ledger_share_s6": 58711600.85497331
},
"V1_02_medium/1/1/matched": {
"status": "scored",
- "ate_4dof": 0.029,
- "ratio": 0.498,
- "rpe_rot_deg": 0.282,
- "nees_yaw": 3.1196205749991686,
- "nees_rp": 2.444168568072167,
- "nees_pos": 0.4680899577036728,
- "yaw_sigma_first": 0.10000125068941469,
- "yaw_sigma_min": 0.017253592823220067,
- "yaw_sigma_ratio": 0.1725,
- "yaw_err_max": 0.04460242887912699,
- "nis": 1.00287,
- "ledger_info_yaw": 159521209.69444683,
- "ledger_share_s6": 159521209.69787452
+ "ate_4dof": 0.0297,
+ "ratio": 0.51,
+ "rpe_rot_deg": 0.3,
+ "nees_yaw": 68.4081128067403,
+ "nees_rp": 2.5756041228737128,
+ "nees_pos": 5.673970267889802,
+ "yaw_sigma_first": 0.10000125068941472,
+ "yaw_sigma_min": 0.015087988737221466,
+ "yaw_sigma_ratio": 0.1509,
+ "yaw_err_max": 0.15663065429484183,
+ "nis": 1.00433,
+ "ledger_info_yaw": 1047049313.8487829,
+ "ledger_share_s6": 1047049313.8455139
},
"V1_02_medium/1/1/default": {
"status": "scored",
- "ate_4dof": 0.0835,
- "ratio": 1.432,
- "rpe_rot_deg": 0.588,
- "nees_yaw": 20.192679891054063,
- "nees_rp": 3.9400649317517584,
- "nees_pos": 3.4069794675250993,
- "yaw_sigma_first": 0.10000125068941469,
- "yaw_sigma_min": 0.019252307145789623,
- "yaw_sigma_ratio": 0.1925,
- "yaw_err_max": 0.1214910340933261,
- "nis": 0.0489708,
- "ledger_info_yaw": 592160106.2749767,
- "ledger_share_s6": 592160106.2757115
+ "ate_4dof": 0.0372,
+ "ratio": 0.638,
+ "rpe_rot_deg": 0.892,
+ "nees_yaw": 0.31546292580962365,
+ "nees_rp": 1.9889952680305711,
+ "nees_pos": 0.25463823892480036,
+ "yaw_sigma_first": 0.10000125068941472,
+ "yaw_sigma_min": 0.05462019898216831,
+ "yaw_sigma_ratio": 0.5462,
+ "yaw_err_max": 0.05086671564666507,
+ "nis": 0.0479126,
+ "ledger_info_yaw": 34809859.107048556,
+ "ledger_share_s6": 34809859.10722664
},
"V1_02_medium/1/2/matched": {
"status": "scored",
- "ate_4dof": 0.0371,
- "ratio": 0.637,
- "rpe_rot_deg": 0.445,
- "nees_yaw": 5.453084023475559,
- "nees_rp": 1.6284227002697884,
- "nees_pos": 1.5274799134573094,
- "yaw_sigma_first": 0.1000012506894147,
- "yaw_sigma_min": 0.024020465071623046,
- "yaw_sigma_ratio": 0.2402,
- "yaw_err_max": 0.0941240097241518,
- "nis": 1.01124,
- "ledger_info_yaw": 53146492.12475925,
- "ledger_share_s6": 53146492.125903964
+ "ate_4dof": 0.0265,
+ "ratio": 0.454,
+ "rpe_rot_deg": 0.356,
+ "nees_yaw": 16.516792853111536,
+ "nees_rp": 2.048492443886388,
+ "nees_pos": 1.9248942699183498,
+ "yaw_sigma_first": 0.10000125068941473,
+ "yaw_sigma_min": 0.015294820377360841,
+ "yaw_sigma_ratio": 0.1529,
+ "yaw_err_max": 0.10902894390169676,
+ "nis": 0.999608,
+ "ledger_info_yaw": 2515059389.410155,
+ "ledger_share_s6": 2515059389.410194
},
"V1_02_medium/1/2/default": {
"status": "scored",
- "ate_4dof": 0.0421,
- "ratio": 0.723,
- "rpe_rot_deg": 0.724,
- "nees_yaw": 0.844263043139033,
- "nees_rp": 1.3350028401319682,
- "nees_pos": 0.35200195911979837,
- "yaw_sigma_first": 0.1000012506894147,
- "yaw_sigma_min": 0.04171556032858866,
- "yaw_sigma_ratio": 0.4172,
- "yaw_err_max": 0.052537641401304215,
- "nis": 0.0477572,
- "ledger_info_yaw": 17274738.170344524,
- "ledger_share_s6": 17274738.170363422
+ "ate_4dof": 0.0338,
+ "ratio": 0.58,
+ "rpe_rot_deg": 0.6,
+ "nees_yaw": 24.695544913370387,
+ "nees_rp": 2.183423760740365,
+ "nees_pos": 6.024786499407295,
+ "yaw_sigma_first": 0.10000125068941473,
+ "yaw_sigma_min": 0.030516932748109713,
+ "yaw_sigma_ratio": 0.3052,
+ "yaw_err_max": 0.1955242528306181,
+ "nis": 0.0478047,
+ "ledger_info_yaw": 48986362.40943415,
+ "ledger_share_s6": 48986362.40949663
},
"V1_02_medium/1/3/matched": {
"status": "scored",
- "ate_4dof": 0.0344,
- "ratio": 0.59,
- "rpe_rot_deg": 0.225,
- "nees_yaw": 2.5462659684719053,
- "nees_rp": 1.8010675156915104,
- "nees_pos": 2.843368321489971,
- "yaw_sigma_first": 0.10000125068941468,
- "yaw_sigma_min": 0.010906635303658891,
- "yaw_sigma_ratio": 0.1091,
- "yaw_err_max": 0.03538982287284909,
- "nis": 1.00725,
- "ledger_info_yaw": 107841201.25139639,
- "ledger_share_s6": 107841201.25108595
+ "ate_4dof": 0.0261,
+ "ratio": 0.447,
+ "rpe_rot_deg": 0.315,
+ "nees_yaw": 14.123563886460705,
+ "nees_rp": 1.6601085555020547,
+ "nees_pos": 1.9725192068149382,
+ "yaw_sigma_first": 0.10000125068941466,
+ "yaw_sigma_min": 0.021691548883976612,
+ "yaw_sigma_ratio": 0.2169,
+ "yaw_err_max": 0.10098119908350114,
+ "nis": 1.00538,
+ "ledger_info_yaw": 38567798.156589806,
+ "ledger_share_s6": 38567798.15719264
},
"V1_02_medium/1/3/default": {
"status": "scored",
- "ate_4dof": 0.0362,
- "ratio": 0.621,
- "rpe_rot_deg": 0.563,
- "nees_yaw": 0.22642742310495384,
- "nees_rp": 0.7051728904490921,
- "nees_pos": 0.12067585060030048,
- "yaw_sigma_first": 0.10000125068941468,
- "yaw_sigma_min": 0.052145139339722556,
- "yaw_sigma_ratio": 0.5214,
- "yaw_err_max": 0.03955628645968099,
- "nis": 0.0480325,
- "ledger_info_yaw": 55856081.09170946,
- "ledger_share_s6": 55856081.09136056
+ "ate_4dof": 0.0311,
+ "ratio": 0.533,
+ "rpe_rot_deg": 0.399,
+ "nees_yaw": 1.2934167738742774,
+ "nees_rp": 1.214851783118833,
+ "nees_pos": 0.371266479246542,
+ "yaw_sigma_first": 0.10000125068941466,
+ "yaw_sigma_min": 0.030832408475682273,
+ "yaw_sigma_ratio": 0.3083,
+ "yaw_err_max": 0.05257031817148607,
+ "nis": 0.048577,
+ "ledger_info_yaw": 7410736.626485181,
+ "ledger_share_s6": 7410736.62654844
},
"V1_02_medium/0.5/1/matched": {
"status": "scored",
- "ate_4dof": 0.0279,
- "ratio": 0.479,
- "rpe_rot_deg": 0.589,
- "nees_yaw": 1.6544222614799973,
- "nees_rp": 2.2986881660849674,
- "nees_pos": 0.6337863272423985,
- "yaw_sigma_first": 0.10000125068941469,
- "yaw_sigma_min": 0.028392032285425143,
- "yaw_sigma_ratio": 0.2839,
- "yaw_err_max": 0.07180002954065769,
- "nis": 1.00306,
- "ledger_info_yaw": 1234704588.2626667,
- "ledger_share_s6": 874021704.9857903
+ "ate_4dof": 0.0218,
+ "ratio": 0.375,
+ "rpe_rot_deg": 0.289,
+ "nees_yaw": 18.352109814906086,
+ "nees_rp": 2.4911096157790675,
+ "nees_pos": 2.6789019141697334,
+ "yaw_sigma_first": 0.10000125068941472,
+ "yaw_sigma_min": 0.019739845285997298,
+ "yaw_sigma_ratio": 0.1974,
+ "yaw_err_max": 0.1500360267917039,
+ "nis": 1.00316,
+ "ledger_info_yaw": 1475389050.5298746,
+ "ledger_share_s6": 1025430404.5702326
},
"V1_02_medium/0.5/1/default": {
"status": "scored",
- "ate_4dof": 0.0353,
- "ratio": 0.605,
- "rpe_rot_deg": 0.834,
- "nees_yaw": 0.3562417125319314,
- "nees_rp": 2.515734041231727,
- "nees_pos": 0.2803652123950202,
- "yaw_sigma_first": 0.10000125068941469,
- "yaw_sigma_min": 0.04962071495190228,
- "yaw_sigma_ratio": 0.4962,
- "yaw_err_max": 0.07319716656318162,
- "nis": 0.0121259,
- "ledger_info_yaw": 27545379.986415185,
- "ledger_share_s6": 22285193.364519536
+ "ate_4dof": 0.0342,
+ "ratio": 0.586,
+ "rpe_rot_deg": 0.773,
+ "nees_yaw": 4.208820676689286,
+ "nees_rp": 2.6878235805730104,
+ "nees_pos": 1.4218817484526722,
+ "yaw_sigma_first": 0.10000125068941472,
+ "yaw_sigma_min": 0.04030949112521348,
+ "yaw_sigma_ratio": 0.4031,
+ "yaw_err_max": 0.1660103404405714,
+ "nis": 0.0121395,
+ "ledger_info_yaw": 41547210.60308283,
+ "ledger_share_s6": 32116641.703297246
},
"V1_02_medium/0.5/2/matched": {
"status": "scored",
- "ate_4dof": 0.024,
- "ratio": 0.412,
- "rpe_rot_deg": 0.878,
- "nees_yaw": 15.129125509745911,
- "nees_rp": 1.5470599341762676,
- "nees_pos": 3.8604714128494395,
- "yaw_sigma_first": 0.1000012506894147,
- "yaw_sigma_min": 0.028816780451427354,
- "yaw_sigma_ratio": 0.2882,
- "yaw_err_max": 0.150848165900498,
- "nis": 0.999824,
- "ledger_info_yaw": 308084905.5186623,
- "ledger_share_s6": 219383650.4485656
+ "ate_4dof": 0.0167,
+ "ratio": 0.286,
+ "rpe_rot_deg": 0.359,
+ "nees_yaw": 3.997770037700464,
+ "nees_rp": 1.533675470923801,
+ "nees_pos": 0.5942335524644741,
+ "yaw_sigma_first": 0.10000125068941473,
+ "yaw_sigma_min": 0.021030275340740045,
+ "yaw_sigma_ratio": 0.2103,
+ "yaw_err_max": 0.11626117971298877,
+ "nis": 1.00006,
+ "ledger_info_yaw": 286925358.46508574,
+ "ledger_share_s6": 132412475.70750113
},
"V1_02_medium/0.5/2/default": {
"status": "scored",
- "ate_4dof": 0.0514,
- "ratio": 0.882,
- "rpe_rot_deg": 0.991,
- "nees_yaw": 0.08679315258310294,
- "nees_rp": 1.5584427861195,
- "nees_pos": 0.07712064584425095,
- "yaw_sigma_first": 0.1000012506894147,
- "yaw_sigma_min": 0.05392452899162289,
- "yaw_sigma_ratio": 0.5392,
- "yaw_err_max": 0.0701547946903022,
- "nis": 0.0141829,
- "ledger_info_yaw": 7375978.141167221,
- "ledger_share_s6": 6788009.363589251
+ "ate_4dof": 0.0505,
+ "ratio": 0.867,
+ "rpe_rot_deg": 1.08,
+ "nees_yaw": 6.601084598913943,
+ "nees_rp": 2.104858104224107,
+ "nees_pos": 2.000991476596963,
+ "yaw_sigma_first": 0.10000125068941473,
+ "yaw_sigma_min": 0.041366087101973065,
+ "yaw_sigma_ratio": 0.4137,
+ "yaw_err_max": 0.1563590437523734,
+ "nis": 0.012179,
+ "ledger_info_yaw": 38561453.283789545,
+ "ledger_share_s6": 29574484.09151142
},
"V1_02_medium/0.5/3/matched": {
"status": "scored",
- "ate_4dof": 0.0296,
- "ratio": 0.508,
- "rpe_rot_deg": 0.657,
- "nees_yaw": 2.7621548747346165,
- "nees_rp": 1.7274058189007944,
- "nees_pos": 0.5523932733641916,
- "yaw_sigma_first": 0.10000125068941468,
- "yaw_sigma_min": 0.024242571597365106,
- "yaw_sigma_ratio": 0.2424,
- "yaw_err_max": 0.07693772591152086,
- "nis": 1.00619,
- "ledger_info_yaw": 882618507.5933361,
- "ledger_share_s6": 660260448.1505322
+ "ate_4dof": 0.0217,
+ "ratio": 0.373,
+ "rpe_rot_deg": 0.461,
+ "nees_yaw": 5.424429013453939,
+ "nees_rp": 1.6100514232979917,
+ "nees_pos": 1.0575427952545593,
+ "yaw_sigma_first": 0.10000125068941466,
+ "yaw_sigma_min": 0.020727320984760943,
+ "yaw_sigma_ratio": 0.2073,
+ "yaw_err_max": 0.11033080806033412,
+ "nis": 1.00625,
+ "ledger_info_yaw": 816964368.4481324,
+ "ledger_share_s6": 513378660.9732304
},
"V1_02_medium/0.5/3/default": {
"status": "scored",
- "ate_4dof": 0.0498,
- "ratio": 0.854,
- "rpe_rot_deg": 1.134,
- "nees_yaw": 1.926038826497069,
- "nees_rp": 1.2117629823903937,
- "nees_pos": 0.5408671364009279,
- "yaw_sigma_first": 0.10000125068941468,
- "yaw_sigma_min": 0.03852445513334371,
- "yaw_sigma_ratio": 0.3852,
- "yaw_err_max": 0.1234523806309527,
- "nis": 0.0122541,
- "ledger_info_yaw": 31523417.39444352,
- "ledger_share_s6": 26640031.852913503
+ "ate_4dof": 0.0464,
+ "ratio": 0.796,
+ "rpe_rot_deg": 0.88,
+ "nees_yaw": 3.3940626349747913,
+ "nees_rp": 1.004498082454721,
+ "nees_pos": 1.0674418999517432,
+ "yaw_sigma_first": 0.10000125068941466,
+ "yaw_sigma_min": 0.036282462988599266,
+ "yaw_sigma_ratio": 0.3628,
+ "yaw_err_max": 0.14741837755728174,
+ "nis": 0.0122204,
+ "ledger_info_yaw": 28509567.080608718,
+ "ledger_share_s6": 22986786.213092487
},
"V1_03_difficult/1/1/matched": {
"status": "scored",
- "ate_4dof": 0.0228,
- "ratio": 0.341,
- "rpe_rot_deg": 0.379,
- "nees_yaw": 26.020711664103906,
- "nees_rp": 3.1277124881062055,
- "nees_pos": 5.638760378262002,
- "yaw_sigma_first": 0.1000012506894147,
- "yaw_sigma_min": 0.0191869444244122,
- "yaw_sigma_ratio": 0.1919,
- "yaw_err_max": 0.11435819456924826,
- "nis": 1.00302,
- "ledger_info_yaw": 1715584005.383729,
- "ledger_share_s6": 1715584005.397909
+ "ate_4dof": 0.024,
+ "ratio": 0.36,
+ "rpe_rot_deg": 0.374,
+ "nees_yaw": 71.64728522403463,
+ "nees_rp": 4.111174312685636,
+ "nees_pos": 6.792486982586141,
+ "yaw_sigma_first": 0.10000125068941469,
+ "yaw_sigma_min": 0.01093233102646939,
+ "yaw_sigma_ratio": 0.1093,
+ "yaw_err_max": 0.116619107572699,
+ "nis": 1.00617,
+ "ledger_info_yaw": 8026999977.265026,
+ "ledger_share_s6": 8026999977.307654
},
"V1_03_difficult/1/1/default": {
"status": "scored",
- "ate_4dof": 0.0254,
- "ratio": 0.381,
- "rpe_rot_deg": 0.307,
- "nees_yaw": 19.87682374200344,
- "nees_rp": 2.080739420239466,
- "nees_pos": 5.51906604582991,
- "yaw_sigma_first": 0.1000012506894147,
- "yaw_sigma_min": 0.029531608446727534,
- "yaw_sigma_ratio": 0.2953,
- "yaw_err_max": 0.1632392175666664,
- "nis": 0.0478407,
- "ledger_info_yaw": 884089605.790051,
- "ledger_share_s6": 884089605.795394
+ "ate_4dof": 0.0463,
+ "ratio": 0.693,
+ "rpe_rot_deg": 0.524,
+ "nees_yaw": 2.871947672654682,
+ "nees_rp": 2.160049512057354,
+ "nees_pos": 0.9689210910478971,
+ "yaw_sigma_first": 0.10000125068941469,
+ "yaw_sigma_min": 0.03747737139330942,
+ "yaw_sigma_ratio": 0.3748,
+ "yaw_err_max": 0.10747327851874683,
+ "nis": 0.0479343,
+ "ledger_info_yaw": 58230832.44144371,
+ "ledger_share_s6": 58230832.44115375
},
"V1_03_difficult/1/2/matched": {
"status": "scored",
- "ate_4dof": 0.0539,
- "ratio": 0.806,
- "rpe_rot_deg": 0.284,
- "nees_yaw": 2.7993714037708557,
- "nees_rp": 1.9973289734183715,
- "nees_pos": 1.0894538225754848,
- "yaw_sigma_first": 0.10000125068941468,
- "yaw_sigma_min": 0.01100760732312817,
- "yaw_sigma_ratio": 0.1101,
- "yaw_err_max": 0.06553300896523732,
- "nis": 1.00153,
- "ledger_info_yaw": 3927207499.2877374,
- "ledger_share_s6": 3927207499.305283
+ "ate_4dof": 0.0207,
+ "ratio": 0.31,
+ "rpe_rot_deg": 0.205,
+ "nees_yaw": 4.6353291489919695,
+ "nees_rp": 1.7226147206510725,
+ "nees_pos": 0.6501318002227883,
+ "yaw_sigma_first": 0.1000012506894147,
+ "yaw_sigma_min": 0.015139822915243021,
+ "yaw_sigma_ratio": 0.1514,
+ "yaw_err_max": 0.07729038832181148,
+ "nis": 0.999777,
+ "ledger_info_yaw": 55996869337.99863,
+ "ledger_share_s6": 55996869338.062874
},
"V1_03_difficult/1/2/default": {
"status": "scored",
- "ate_4dof": 0.046,
- "ratio": 0.688,
- "rpe_rot_deg": 0.487,
- "nees_yaw": 0.03726737220685129,
- "nees_rp": 0.8349236547932495,
- "nees_pos": 0.9739574282033506,
- "yaw_sigma_first": 0.10000125068941468,
- "yaw_sigma_min": 0.03696684629849992,
- "yaw_sigma_ratio": 0.3697,
- "yaw_err_max": 0.03370323840461925,
- "nis": 0.0477064,
- "ledger_info_yaw": 446393799.53275734,
- "ledger_share_s6": 446393799.5346484
+ "ate_4dof": 0.0602,
+ "ratio": 0.902,
+ "rpe_rot_deg": 0.65,
+ "nees_yaw": 11.152421832586189,
+ "nees_rp": 2.556477631241017,
+ "nees_pos": 2.3773134568865895,
+ "yaw_sigma_first": 0.1000012506894147,
+ "yaw_sigma_min": 0.009600098695041557,
+ "yaw_sigma_ratio": 0.096,
+ "yaw_err_max": 0.1295732494320981,
+ "nis": 0.0485793,
+ "ledger_info_yaw": 5844240666.646142,
+ "ledger_share_s6": 5844240666.651881
},
"V1_03_difficult/1/3/matched": {
"status": "scored",
- "ate_4dof": 0.0195,
- "ratio": 0.292,
- "rpe_rot_deg": 0.207,
- "nees_yaw": 0.586813588755702,
- "nees_rp": 2.702816400552285,
- "nees_pos": 0.5668315708598441,
+ "ate_4dof": 0.0202,
+ "ratio": 0.302,
+ "rpe_rot_deg": 0.286,
+ "nees_yaw": 50.557097501705016,
+ "nees_rp": 2.864661158152819,
+ "nees_pos": 6.323516077188598,
"yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.011905956768664601,
- "yaw_sigma_ratio": 0.1191,
- "yaw_err_max": 0.01270479899804988,
- "nis": 1.00346,
- "ledger_info_yaw": 150252276.90411967,
- "ledger_share_s6": 150252276.90193555
+ "yaw_sigma_min": 0.01300109551046586,
+ "yaw_sigma_ratio": 0.13,
+ "yaw_err_max": 0.13826732428295854,
+ "nis": 1.00315,
+ "ledger_info_yaw": 4556680177.894437,
+ "ledger_share_s6": 4556680177.913166
},
"V1_03_difficult/1/3/default": {
"status": "scored",
- "ate_4dof": 0.0347,
- "ratio": 0.519,
- "rpe_rot_deg": 0.415,
- "nees_yaw": 1.2760491047800444,
- "nees_rp": 1.8947044266096142,
- "nees_pos": 0.5264933428535045,
+ "ate_4dof": 0.0568,
+ "ratio": 0.85,
+ "rpe_rot_deg": 0.363,
+ "nees_yaw": 0.36639414535214276,
+ "nees_rp": 2.026100204326056,
+ "nees_pos": 1.3410821027087563,
"yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.046647828419943825,
- "yaw_sigma_ratio": 0.4665,
- "yaw_err_max": 0.08083021736088086,
- "nis": 0.0479224,
- "ledger_info_yaw": 13582705.594379866,
- "ledger_share_s6": 13582705.59437174
+ "yaw_sigma_min": 0.02823098589767569,
+ "yaw_sigma_ratio": 0.2823,
+ "yaw_err_max": 0.035326245584859,
+ "nis": 0.0479502,
+ "ledger_info_yaw": 509542121.28483045,
+ "ledger_share_s6": 509542121.2852781
},
"V1_03_difficult/0.5/1/matched": {
"status": "scored",
- "ate_4dof": 0.0295,
- "ratio": 0.442,
- "rpe_rot_deg": 0.818,
- "nees_yaw": 7.644409158805222,
- "nees_rp": 4.049182149204656,
- "nees_pos": 1.5367103786427507,
- "yaw_sigma_first": 0.1000012506894147,
- "yaw_sigma_min": 0.022739030327333747,
- "yaw_sigma_ratio": 0.2274,
- "yaw_err_max": 0.09448968341929408,
- "nis": 1.00245,
- "ledger_info_yaw": 12469501580.657532,
- "ledger_share_s6": 6159031758.026282
+ "ate_4dof": 0.0166,
+ "ratio": 0.249,
+ "rpe_rot_deg": 0.384,
+ "nees_yaw": 6.2820104459093535,
+ "nees_rp": 4.024035801451964,
+ "nees_pos": 0.7579402900088393,
+ "yaw_sigma_first": 0.10000125068941469,
+ "yaw_sigma_min": 0.015535338926130658,
+ "yaw_sigma_ratio": 0.1554,
+ "yaw_err_max": 0.05611927402970084,
+ "nis": 1.00235,
+ "ledger_info_yaw": 24266906391.697296,
+ "ledger_share_s6": -10683345577.499369
},
"V1_03_difficult/0.5/1/default": {
"status": "scored",
- "ate_4dof": 0.0341,
- "ratio": 0.511,
- "rpe_rot_deg": 0.746,
- "nees_yaw": 0.13625192273077233,
- "nees_rp": 2.0717167988378447,
- "nees_pos": 0.18703808491511592,
- "yaw_sigma_first": 0.1000012506894147,
- "yaw_sigma_min": 0.06508358242971875,
- "yaw_sigma_ratio": 0.6508,
- "yaw_err_max": 0.06495143196690706,
- "nis": 0.0120763,
- "ledger_info_yaw": 12209887.053580327,
- "ledger_share_s6": 46342.981965086
+ "ate_4dof": 0.0297,
+ "ratio": 0.445,
+ "rpe_rot_deg": 0.522,
+ "nees_yaw": 0.5983713349144281,
+ "nees_rp": 2.299071151917543,
+ "nees_pos": 0.3226332515410065,
+ "yaw_sigma_first": 0.10000125068941469,
+ "yaw_sigma_min": 0.045882625025870385,
+ "yaw_sigma_ratio": 0.4588,
+ "yaw_err_max": 0.05324873400741287,
+ "nis": 0.0120956,
+ "ledger_info_yaw": 15662839.473670742,
+ "ledger_share_s6": -7048344.085185562
},
"V1_03_difficult/0.5/2/matched": {
"status": "scored",
- "ate_4dof": 0.0277,
- "ratio": 0.415,
- "rpe_rot_deg": 0.699,
- "nees_yaw": 0.5823463813564803,
- "nees_rp": 1.5373011692661052,
- "nees_pos": 0.2526548137934055,
- "yaw_sigma_first": 0.10000125068941468,
- "yaw_sigma_min": 0.026659246174714307,
- "yaw_sigma_ratio": 0.2666,
- "yaw_err_max": 0.053859604673164085,
- "nis": 0.998936,
- "ledger_info_yaw": 3404192301.622328,
- "ledger_share_s6": 1594509548.8009872
+ "ate_4dof": 0.0171,
+ "ratio": 0.257,
+ "rpe_rot_deg": 0.394,
+ "nees_yaw": 18.161976564727834,
+ "nees_rp": 1.536910178727021,
+ "nees_pos": 2.3176062538813866,
+ "yaw_sigma_first": 0.1000012506894147,
+ "yaw_sigma_min": 0.019168212518263724,
+ "yaw_sigma_ratio": 0.1917,
+ "yaw_err_max": 0.127460849638856,
+ "nis": 0.999066,
+ "ledger_info_yaw": 24965452868.78123,
+ "ledger_share_s6": -2459449337.7937965
},
"V1_03_difficult/0.5/2/default": {
"status": "scored",
- "ate_4dof": 0.0277,
- "ratio": 0.414,
- "rpe_rot_deg": 0.668,
- "nees_yaw": 0.02217156902804229,
- "nees_rp": 0.8046647960133595,
- "nees_pos": 0.11702907525822988,
- "yaw_sigma_first": 0.10000125068941468,
- "yaw_sigma_min": 0.07017719990686265,
- "yaw_sigma_ratio": 0.7018,
- "yaw_err_max": 0.03152595007860625,
- "nis": 0.0120536,
- "ledger_info_yaw": 19090822.08828791,
- "ledger_share_s6": 8651394.01125396
+ "ate_4dof": 0.0293,
+ "ratio": 0.439,
+ "rpe_rot_deg": 0.477,
+ "nees_yaw": 7.267089051868656,
+ "nees_rp": 1.111776150667874,
+ "nees_pos": 2.4677898108756136,
+ "yaw_sigma_first": 0.1000012506894147,
+ "yaw_sigma_min": 0.04839373801176506,
+ "yaw_sigma_ratio": 0.4839,
+ "yaw_err_max": 0.17229964872209114,
+ "nis": 0.0120842,
+ "ledger_info_yaw": 5631747353.318376,
+ "ledger_share_s6": 1459885747.0412626
},
"V1_03_difficult/0.5/3/matched": {
"status": "scored",
- "ate_4dof": 0.0142,
- "ratio": 0.212,
- "rpe_rot_deg": 0.424,
- "nees_yaw": 19.29844025880208,
- "nees_rp": 2.581863095007469,
- "nees_pos": 2.5870143458903687,
+ "ate_4dof": 0.0135,
+ "ratio": 0.203,
+ "rpe_rot_deg": 0.375,
+ "nees_yaw": 2.1135434847258905,
+ "nees_rp": 2.720822121392351,
+ "nees_pos": 0.24937952340235284,
"yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.018140698757945183,
- "yaw_sigma_ratio": 0.1814,
- "yaw_err_max": 0.11072452384628341,
- "nis": 1.00341,
- "ledger_info_yaw": 701622567.2973316,
- "ledger_share_s6": 43917863.13708817
+ "yaw_sigma_min": 0.015597549180728927,
+ "yaw_sigma_ratio": 0.156,
+ "yaw_err_max": 0.07774957960324341,
+ "nis": 1.00348,
+ "ledger_info_yaw": 1268428903.4090738,
+ "ledger_share_s6": -262448056.50613397
},
"V1_03_difficult/0.5/3/default": {
"status": "scored",
- "ate_4dof": 0.0353,
- "ratio": 0.529,
- "rpe_rot_deg": 0.387,
- "nees_yaw": 0.033031107669617685,
- "nees_rp": 1.9799332942538759,
- "nees_pos": 0.0780864069711533,
+ "ate_4dof": 0.0406,
+ "ratio": 0.608,
+ "rpe_rot_deg": 0.433,
+ "nees_yaw": 4.218609192688085,
+ "nees_rp": 2.16710752202191,
+ "nees_pos": 0.8379230249241888,
"yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.04863332036221474,
- "yaw_sigma_ratio": 0.4863,
- "yaw_err_max": 0.04213459113130332,
- "nis": 0.0121268,
- "ledger_info_yaw": 8195196.403132529,
- "ledger_share_s6": -460599.8737669339
+ "yaw_sigma_min": 0.031052289424049306,
+ "yaw_sigma_ratio": 0.3105,
+ "yaw_err_max": 0.1012472966128459,
+ "nis": 0.0121422,
+ "ledger_info_yaw": 39407822.32885517,
+ "ledger_share_s6": 4039675.578363329
},
"V2_01_easy/1/1/matched": {
"status": "scored",
- "ate_4dof": 0.0249,
- "ratio": 0.196,
- "rpe_rot_deg": 0.333,
- "nees_yaw": 1.1822959088070433,
- "nees_rp": 1.6386107026741865,
- "nees_pos": 0.40521437406985117,
- "yaw_sigma_first": 0.10003124796018363,
- "yaw_sigma_min": 0.01925211388883723,
- "yaw_sigma_ratio": 0.1925,
- "yaw_err_max": 0.04078760293644753,
- "nis": 1.0022,
- "ledger_info_yaw": 449781918635.59937,
- "ledger_share_s6": 449781918635.8579
+ "ate_4dof": 0.0209,
+ "ratio": 0.165,
+ "rpe_rot_deg": 0.815,
+ "nees_yaw": 11.240893533974305,
+ "nees_rp": 3.7701374922335287,
+ "nees_pos": 1.0893091068931116,
+ "yaw_sigma_first": 0.10003124796018369,
+ "yaw_sigma_min": 0.00998257830873048,
+ "yaw_sigma_ratio": 0.0998,
+ "yaw_err_max": 0.06542487603643889,
+ "nis": 1.0098,
+ "ledger_info_yaw": 6407869350.691273,
+ "ledger_share_s6": 6407869350.590326
},
"V2_01_easy/1/1/default": {
"status": "scored",
- "ate_4dof": 0.0298,
- "ratio": 0.235,
- "rpe_rot_deg": 0.884,
- "nees_yaw": 0.10087251181430286,
- "nees_rp": 1.7368295582876814,
- "nees_pos": 0.38463239661794624,
- "yaw_sigma_first": 0.10003124796018363,
- "yaw_sigma_min": 0.04941221712782,
- "yaw_sigma_ratio": 0.494,
- "yaw_err_max": 0.048489322833372345,
- "nis": 0.0478634,
- "ledger_info_yaw": 362158048.93657225,
- "ledger_share_s6": 362158048.93729204
+ "ate_4dof": 0.0408,
+ "ratio": 0.321,
+ "rpe_rot_deg": 0.909,
+ "nees_yaw": 73.11277927368789,
+ "nees_rp": 3.7205111290037896,
+ "nees_pos": 8.05137390974486,
+ "yaw_sigma_first": 0.10003124796018369,
+ "yaw_sigma_min": 0.011697757762797675,
+ "yaw_sigma_ratio": 0.1169,
+ "yaw_err_max": 0.20405880301692972,
+ "nis": 0.0487854,
+ "ledger_info_yaw": 35124261785.88357,
+ "ledger_share_s6": 35124261785.93763
},
"V2_01_easy/1/2/matched": {
"status": "scored",
- "ate_4dof": 0.0301,
- "ratio": 0.237,
- "rpe_rot_deg": 0.77,
- "nees_yaw": 0.2013990736447985,
- "nees_rp": 1.3960271520890224,
- "nees_pos": 0.4568167071503049,
- "yaw_sigma_first": 0.10003124796016971,
- "yaw_sigma_min": 0.021439557522266253,
- "yaw_sigma_ratio": 0.2143,
- "yaw_err_max": 0.02947503784208602,
- "nis": 1.00122,
- "ledger_info_yaw": 6117924425.0717,
- "ledger_share_s6": 6117924425.077955
+ "ate_4dof": 0.0311,
+ "ratio": 0.245,
+ "rpe_rot_deg": 0.638,
+ "nees_yaw": 280.4161003956329,
+ "nees_rp": 2.046974338612236,
+ "nees_pos": 22.217887914747838,
+ "yaw_sigma_first": 0.10003124796016974,
+ "yaw_sigma_min": 0.010420354073190616,
+ "yaw_sigma_ratio": 0.1042,
+ "yaw_err_max": 0.22026737218703338,
+ "nis": 1.00648,
+ "ledger_info_yaw": 21544653439.29501,
+ "ledger_share_s6": 21544653439.27894
},
"V2_01_easy/1/2/default": {
"status": "scored",
- "ate_4dof": 0.0376,
- "ratio": 0.296,
- "rpe_rot_deg": 0.872,
- "nees_yaw": 0.3273968826485229,
- "nees_rp": 1.0678538516757439,
- "nees_pos": 0.1976069717178499,
- "yaw_sigma_first": 0.10003124796016971,
- "yaw_sigma_min": 0.05715520739899881,
- "yaw_sigma_ratio": 0.5714,
- "yaw_err_max": 0.05912765469369294,
- "nis": 0.0477128,
- "ledger_info_yaw": 112148349.93750624,
- "ledger_share_s6": 112148349.93746513
+ "ate_4dof": 0.0487,
+ "ratio": 0.384,
+ "rpe_rot_deg": 1.037,
+ "nees_yaw": 4.45427200700438,
+ "nees_rp": 3.433187045844662,
+ "nees_pos": 1.6086292490558958,
+ "yaw_sigma_first": 0.10003124796016974,
+ "yaw_sigma_min": 0.010472523744622085,
+ "yaw_sigma_ratio": 0.1047,
+ "yaw_err_max": 0.07383862928661977,
+ "nis": 0.0483085,
+ "ledger_info_yaw": 299278522.7742225,
+ "ledger_share_s6": 299278522.77375644
},
"V2_01_easy/1/3/matched": {
"status": "scored",
- "ate_4dof": 0.0375,
- "ratio": 0.295,
- "rpe_rot_deg": 0.556,
- "nees_yaw": 0.4473670780461336,
- "nees_rp": 1.7829571533962136,
- "nees_pos": 0.43564514280410555,
+ "ate_4dof": 0.0307,
+ "ratio": 0.242,
+ "rpe_rot_deg": 0.433,
+ "nees_yaw": 8.479715612293324,
+ "nees_rp": 1.912556162382845,
+ "nees_pos": 0.9407367799652266,
"yaw_sigma_first": 0.10003124796016802,
- "yaw_sigma_min": 0.026398288327401112,
- "yaw_sigma_ratio": 0.2639,
- "yaw_err_max": 0.07009926902121827,
- "nis": 1.00447,
- "ledger_info_yaw": 9832720674.537436,
- "ledger_share_s6": 9832720674.571424
+ "yaw_sigma_min": 0.01101140776111771,
+ "yaw_sigma_ratio": 0.1101,
+ "yaw_err_max": 0.06161890903649371,
+ "nis": 1.00772,
+ "ledger_info_yaw": 27057742845.258587,
+ "ledger_share_s6": 27057742845.424442
},
"V2_01_easy/1/3/default": {
"status": "scored",
- "ate_4dof": 0.0409,
- "ratio": 0.323,
- "rpe_rot_deg": 0.617,
- "nees_yaw": 0.5620604860567961,
- "nees_rp": 2.173684558892366,
- "nees_pos": 0.2535068987449784,
+ "ate_4dof": 0.0672,
+ "ratio": 0.53,
+ "rpe_rot_deg": 0.833,
+ "nees_yaw": 39.69557151405326,
+ "nees_rp": 2.4209015128748907,
+ "nees_pos": 5.182616388074564,
"yaw_sigma_first": 0.10003124796016802,
- "yaw_sigma_min": 0.02201814078574119,
- "yaw_sigma_ratio": 0.2201,
- "yaw_err_max": 0.03016395124506139,
- "nis": 0.0480131,
- "ledger_info_yaw": 1657267769.198126,
- "ledger_share_s6": 1657267769.2013295
+ "yaw_sigma_min": 0.01495046519012324,
+ "yaw_sigma_ratio": 0.1495,
+ "yaw_err_max": 0.1412197288145764,
+ "nis": 0.0488169,
+ "ledger_info_yaw": 66948241.715062015,
+ "ledger_share_s6": 66948241.714796856
},
"V2_01_easy/0.5/1/matched": {
"status": "scored",
- "ate_4dof": 0.022,
- "ratio": 0.174,
- "rpe_rot_deg": 0.886,
- "nees_yaw": 2.1288557061645093,
- "nees_rp": 1.8165302792300835,
- "nees_pos": 0.8555941310469682,
- "yaw_sigma_first": 0.10003124796018363,
- "yaw_sigma_min": 0.023809867564513903,
- "yaw_sigma_ratio": 0.238,
- "yaw_err_max": 0.0737054022217345,
- "nis": 1.00221,
- "ledger_info_yaw": 7078301595.523178,
- "ledger_share_s6": 3057755211.439608
+ "ate_4dof": 0.0164,
+ "ratio": 0.129,
+ "rpe_rot_deg": 0.385,
+ "nees_yaw": 2.585422099139134,
+ "nees_rp": 2.314701590255161,
+ "nees_pos": 0.41871655319744067,
+ "yaw_sigma_first": 0.10003124796018369,
+ "yaw_sigma_min": 0.011421129292529296,
+ "yaw_sigma_ratio": 0.1142,
+ "yaw_err_max": 0.03497409925174923,
+ "nis": 1.0025,
+ "ledger_info_yaw": 8657361978.251274,
+ "ledger_share_s6": 10677839630.315783
},
"V2_01_easy/0.5/1/default": {
"status": "scored",
- "ate_4dof": 0.0327,
- "ratio": 0.258,
- "rpe_rot_deg": 1.174,
- "nees_yaw": 0.22637757071376158,
- "nees_rp": 1.806508883430062,
- "nees_pos": 0.1348489405722461,
- "yaw_sigma_first": 0.10003124796018363,
- "yaw_sigma_min": 0.04909833024368156,
- "yaw_sigma_ratio": 0.4908,
- "yaw_err_max": 0.07736943265139598,
- "nis": 0.0121442,
- "ledger_info_yaw": 203817079.10281566,
- "ledger_share_s6": 231517058.9238448
+ "ate_4dof": 0.036,
+ "ratio": 0.284,
+ "rpe_rot_deg": 1.403,
+ "nees_yaw": 1.7185004311388152,
+ "nees_rp": 4.7904249822557805,
+ "nees_pos": 0.22915979787139362,
+ "yaw_sigma_first": 0.10003124796018369,
+ "yaw_sigma_min": 0.014511403418966612,
+ "yaw_sigma_ratio": 0.1451,
+ "yaw_err_max": 0.07737546995671062,
+ "nis": 0.0123722,
+ "ledger_info_yaw": 394081464.7323708,
+ "ledger_share_s6": 370099181.98806053
},
"V2_01_easy/0.5/2/matched": {
"status": "scored",
- "ate_4dof": 0.0246,
- "ratio": 0.194,
- "rpe_rot_deg": 0.892,
- "nees_yaw": 10.384045844254526,
- "nees_rp": 1.2131321355693694,
- "nees_pos": 3.0479235010412165,
- "yaw_sigma_first": 0.10003124796016971,
- "yaw_sigma_min": 0.021922915874585155,
- "yaw_sigma_ratio": 0.2192,
- "yaw_err_max": 0.14536629972960177,
- "nis": 1.00046,
- "ledger_info_yaw": 184348867412.388,
- "ledger_share_s6": -24532885812.407017
+ "ate_4dof": 0.0199,
+ "ratio": 0.157,
+ "rpe_rot_deg": 0.551,
+ "nees_yaw": 5.398933292921793,
+ "nees_rp": 1.259214955260205,
+ "nees_pos": 0.7622879221070209,
+ "yaw_sigma_first": 0.10003124796016974,
+ "yaw_sigma_min": 0.016014809079085604,
+ "yaw_sigma_ratio": 0.1601,
+ "yaw_err_max": 0.04567261820622868,
+ "nis": 1.00085,
+ "ledger_info_yaw": 272649134622.41348,
+ "ledger_share_s6": 630468753007.166
},
"V2_01_easy/0.5/2/default": {
"status": "scored",
- "ate_4dof": 0.0321,
- "ratio": 0.253,
- "rpe_rot_deg": 0.726,
- "nees_yaw": 1.2711201781757495,
- "nees_rp": 0.94779067571997,
- "nees_pos": 0.6350064769365686,
- "yaw_sigma_first": 0.10003124796016971,
- "yaw_sigma_min": 0.04538910708927404,
- "yaw_sigma_ratio": 0.4537,
- "yaw_err_max": 0.08686576269994879,
- "nis": 0.0121238,
- "ledger_info_yaw": 80584788.8497221,
- "ledger_share_s6": 26083487.699021995
+ "ate_4dof": 0.0382,
+ "ratio": 0.301,
+ "rpe_rot_deg": 1.464,
+ "nees_yaw": 6.358074613758007,
+ "nees_rp": 3.728039598503503,
+ "nees_pos": 0.698185117516936,
+ "yaw_sigma_first": 0.10003124796016974,
+ "yaw_sigma_min": 0.015854087585742964,
+ "yaw_sigma_ratio": 0.1585,
+ "yaw_err_max": 0.11901872101130195,
+ "nis": 0.0123704,
+ "ledger_info_yaw": 387084582.87179595,
+ "ledger_share_s6": 440496244.37730396
},
"V2_01_easy/0.5/3/matched": {
"status": "scored",
- "ate_4dof": 0.0287,
- "ratio": 0.227,
- "rpe_rot_deg": 1.33,
- "nees_yaw": 28.008915181331844,
- "nees_rp": 1.821002196967001,
- "nees_pos": 8.123548773203344,
+ "ate_4dof": 0.023,
+ "ratio": 0.181,
+ "rpe_rot_deg": 0.723,
+ "nees_yaw": 9.517363815617303,
+ "nees_rp": 2.0118359202298257,
+ "nees_pos": 1.2722133617362925,
"yaw_sigma_first": 0.10003124796016802,
- "yaw_sigma_min": 0.022547204489609448,
- "yaw_sigma_ratio": 0.2254,
- "yaw_err_max": 0.21124463151183967,
- "nis": 1.00499,
- "ledger_info_yaw": 16552255782.652927,
- "ledger_share_s6": 1608765715.16179
+ "yaw_sigma_min": 0.013572286124469141,
+ "yaw_sigma_ratio": 0.1357,
+ "yaw_err_max": 0.07439559428740937,
+ "nis": 1.006,
+ "ledger_info_yaw": 24712651354.062645,
+ "ledger_share_s6": 28787892831.62122
},
"V2_01_easy/0.5/3/default": {
"status": "scored",
- "ate_4dof": 0.046,
- "ratio": 0.363,
- "rpe_rot_deg": 1.299,
- "nees_yaw": 0.08321365312127126,
- "nees_rp": 1.9567171056054409,
- "nees_pos": 0.24234920220654763,
+ "ate_4dof": 0.041,
+ "ratio": 0.323,
+ "rpe_rot_deg": 1.616,
+ "nees_yaw": 0.8600972766006795,
+ "nees_rp": 4.0695383890438315,
+ "nees_pos": 0.3256351236953413,
"yaw_sigma_first": 0.10003124796016802,
- "yaw_sigma_min": 0.04405537651884434,
- "yaw_sigma_ratio": 0.4404,
- "yaw_err_max": 0.045364639679682316,
- "nis": 0.0122287,
- "ledger_info_yaw": 1194449763.0731459,
- "ledger_share_s6": 167658287.40165862
+ "yaw_sigma_min": 0.014135766974874205,
+ "yaw_sigma_ratio": 0.1413,
+ "yaw_err_max": 0.055233152544670504,
+ "nis": 0.0124802,
+ "ledger_info_yaw": 1936902843.8634672,
+ "ledger_share_s6": 2015865114.1303031
},
"V2_02_medium/1/1/matched": {
"status": "scored",
- "ate_4dof": 0.0166,
- "ratio": 0.228,
- "rpe_rot_deg": 0.266,
- "nees_yaw": 11.59331250846095,
- "nees_rp": 2.530774589059129,
- "nees_pos": 1.2553961416070007,
- "yaw_sigma_first": 0.10004499291706767,
- "yaw_sigma_min": 0.014960363241488693,
- "yaw_sigma_ratio": 0.1495,
- "yaw_err_max": 0.07582197014173654,
- "nis": 1.00098,
- "ledger_info_yaw": 107658009305.16829,
- "ledger_share_s6": 107658009304.74838
+ "ate_4dof": 0.0176,
+ "ratio": 0.242,
+ "rpe_rot_deg": 0.369,
+ "nees_yaw": 80.83007390833046,
+ "nees_rp": 4.131920470478102,
+ "nees_pos": 5.742906597580426,
+ "yaw_sigma_first": 0.10004499291706773,
+ "yaw_sigma_min": 0.012690430476665281,
+ "yaw_sigma_ratio": 0.1268,
+ "yaw_err_max": 0.19092514006569117,
+ "nis": 1.0019,
+ "ledger_info_yaw": 1128707026965.1172,
+ "ledger_share_s6": 1128707026962.275
},
"V2_02_medium/1/1/default": {
"status": "scored",
- "ate_4dof": 0.0321,
- "ratio": 0.44,
- "rpe_rot_deg": 0.714,
- "nees_yaw": 0.034481444374026275,
- "nees_rp": 1.707906275353542,
- "nees_pos": 0.08601736112599587,
- "yaw_sigma_first": 0.10004499291706767,
- "yaw_sigma_min": 0.06903243386462429,
- "yaw_sigma_ratio": 0.69,
- "yaw_err_max": 0.02603524443997325,
- "nis": 0.0477198,
- "ledger_info_yaw": 680601242.2966337,
- "ledger_share_s6": 680601242.2959167
+ "ate_4dof": 0.0708,
+ "ratio": 0.972,
+ "rpe_rot_deg": 1.109,
+ "nees_yaw": 1.300539362962507,
+ "nees_rp": 6.481708692252956,
+ "nees_pos": 1.7583137049784063,
+ "yaw_sigma_first": 0.10004499291706773,
+ "yaw_sigma_min": 0.022849245163970435,
+ "yaw_sigma_ratio": 0.2284,
+ "yaw_err_max": 0.11709974891750824,
+ "nis": 0.048152,
+ "ledger_info_yaw": 10225976186.030977,
+ "ledger_share_s6": 10225976186.030268
},
"V2_02_medium/1/2/matched": {
"status": "scored",
- "ate_4dof": 0.0329,
- "ratio": 0.452,
- "rpe_rot_deg": 0.669,
- "nees_yaw": 0.2780412187360738,
- "nees_rp": 1.3726301636779534,
- "nees_pos": 0.15119370080112407,
+ "ate_4dof": 0.0255,
+ "ratio": 0.351,
+ "rpe_rot_deg": 0.381,
+ "nees_yaw": 7.759385594966135,
+ "nees_rp": 2.7726585785489863,
+ "nees_pos": 0.8018157531734557,
"yaw_sigma_first": 0.1000449929170029,
- "yaw_sigma_min": 0.022172452225486958,
- "yaw_sigma_ratio": 0.2216,
- "yaw_err_max": 0.03818784449792656,
- "nis": 1.00131,
- "ledger_info_yaw": 9331542871.578836,
- "ledger_share_s6": 9331542871.567389
+ "yaw_sigma_min": 0.01272423183355341,
+ "yaw_sigma_ratio": 0.1272,
+ "yaw_err_max": 0.06026778380210045,
+ "nis": 1.00279,
+ "ledger_info_yaw": 18563052836.660828,
+ "ledger_share_s6": 18563052836.616165
},
"V2_02_medium/1/2/default": {
"status": "scored",
- "ate_4dof": 0.035,
- "ratio": 0.481,
- "rpe_rot_deg": 0.652,
- "nees_yaw": 0.08343475201492508,
- "nees_rp": 1.0699539813886711,
- "nees_pos": 0.1473191020083439,
+ "ate_4dof": 0.0575,
+ "ratio": 0.79,
+ "rpe_rot_deg": 0.769,
+ "nees_yaw": 0.2432319167333129,
+ "nees_rp": 4.648530659161306,
+ "nees_pos": 1.6362373146917895,
"yaw_sigma_first": 0.1000449929170029,
- "yaw_sigma_min": 0.06860469623925547,
- "yaw_sigma_ratio": 0.6857,
- "yaw_err_max": 0.03728624449254736,
- "nis": 0.0477389,
- "ledger_info_yaw": 287203415.5178894,
- "ledger_share_s6": 287203415.51723516
+ "yaw_sigma_min": 0.025203872259899236,
+ "yaw_sigma_ratio": 0.2519,
+ "yaw_err_max": 0.0709113163911219,
+ "nis": 0.0480881,
+ "ledger_info_yaw": 526387829.6945176,
+ "ledger_share_s6": 526387829.6942683
},
"V2_02_medium/1/3/matched": {
"status": "scored",
- "ate_4dof": 0.02,
- "ratio": 0.274,
- "rpe_rot_deg": 0.442,
- "nees_yaw": 52.45954988336716,
- "nees_rp": 2.0462014379762543,
- "nees_pos": 9.42125048084932,
- "yaw_sigma_first": 0.1000449929169597,
- "yaw_sigma_min": 0.021069807375784302,
- "yaw_sigma_ratio": 0.2106,
- "yaw_err_max": 0.17492591077184402,
- "nis": 1.00226,
- "ledger_info_yaw": 23343438217.24881,
- "ledger_share_s6": 23343438217.22217
+ "ate_4dof": 0.0207,
+ "ratio": 0.284,
+ "rpe_rot_deg": 0.263,
+ "nees_yaw": 2.8023493770250263,
+ "nees_rp": 2.3334514597011866,
+ "nees_pos": 0.2130559316009389,
+ "yaw_sigma_first": 0.10004499291695969,
+ "yaw_sigma_min": 0.008666888051879134,
+ "yaw_sigma_ratio": 0.0866,
+ "yaw_err_max": 0.07009342791852248,
+ "nis": 1.0063,
+ "ledger_info_yaw": 370764004.84995615,
+ "ledger_share_s6": 370764004.85668945
},
"V2_02_medium/1/3/default": {
"status": "scored",
- "ate_4dof": 0.0326,
- "ratio": 0.448,
- "rpe_rot_deg": 0.635,
- "nees_yaw": 1.002580648100769,
- "nees_rp": 1.7605792435266217,
- "nees_pos": 1.260147349037769,
- "yaw_sigma_first": 0.1000449929169597,
- "yaw_sigma_min": 0.06147220495701512,
- "yaw_sigma_ratio": 0.6144,
- "yaw_err_max": 0.0836152390855533,
- "nis": 0.0478828,
- "ledger_info_yaw": 288490582.3772584,
- "ledger_share_s6": 288490582.3763509
+ "ate_4dof": 0.0438,
+ "ratio": 0.601,
+ "rpe_rot_deg": 0.874,
+ "nees_yaw": 4.001948220886447,
+ "nees_rp": 3.0976329274603005,
+ "nees_pos": 0.7729655458209473,
+ "yaw_sigma_first": 0.10004499291695969,
+ "yaw_sigma_min": 0.023821086303574743,
+ "yaw_sigma_ratio": 0.2381,
+ "yaw_err_max": 0.09844694414832084,
+ "nis": 0.0480926,
+ "ledger_info_yaw": 506671948.0206532,
+ "ledger_share_s6": 506671948.0210474
},
"V2_02_medium/0.5/1/matched": {
"status": "scored",
- "ate_4dof": 0.0179,
- "ratio": 0.245,
- "rpe_rot_deg": 0.708,
- "nees_yaw": 8.217011222533309,
- "nees_rp": 2.0559898471026017,
- "nees_pos": 1.8808750307769753,
- "yaw_sigma_first": 0.10004499291706767,
- "yaw_sigma_min": 0.023391720414151676,
- "yaw_sigma_ratio": 0.2338,
- "yaw_err_max": 0.11838129750200853,
- "nis": 1.00084,
- "ledger_info_yaw": 52364020011.30095,
- "ledger_share_s6": 19347384626.375385
+ "ate_4dof": 0.0127,
+ "ratio": 0.174,
+ "rpe_rot_deg": 0.355,
+ "nees_yaw": 0.6023367274481884,
+ "nees_rp": 2.5252884979961485,
+ "nees_pos": 0.07871121074934348,
+ "yaw_sigma_first": 0.10004499291706773,
+ "yaw_sigma_min": 0.015291003340520292,
+ "yaw_sigma_ratio": 0.1528,
+ "yaw_err_max": 0.046311387940588346,
+ "nis": 1.00111,
+ "ledger_info_yaw": 27066349295.660427,
+ "ledger_share_s6": 24117522981.03949
},
"V2_02_medium/0.5/1/default": {
"status": "scored",
- "ate_4dof": 0.0284,
- "ratio": 0.39,
- "rpe_rot_deg": 0.642,
- "nees_yaw": 0.03299600572955008,
- "nees_rp": 1.523647356410312,
- "nees_pos": 0.09455460747401324,
- "yaw_sigma_first": 0.10004499291706767,
- "yaw_sigma_min": 0.05947272509570226,
- "yaw_sigma_ratio": 0.5945,
- "yaw_err_max": 0.07313986627601778,
- "nis": 0.01209,
- "ledger_info_yaw": 219294494.60837144,
- "ledger_share_s6": 91736865.54656973
+ "ate_4dof": 0.0489,
+ "ratio": 0.672,
+ "rpe_rot_deg": 1.049,
+ "nees_yaw": 7.792742973306674,
+ "nees_rp": 4.241106963431272,
+ "nees_pos": 2.9368114762931277,
+ "yaw_sigma_first": 0.10004499291706773,
+ "yaw_sigma_min": 0.03210209828385767,
+ "yaw_sigma_ratio": 0.3209,
+ "yaw_err_max": 0.1575939082011954,
+ "nis": 0.0121793,
+ "ledger_info_yaw": 229102290.63856876,
+ "ledger_share_s6": 218467513.49137485
},
"V2_02_medium/0.5/2/matched": {
"status": "scored",
- "ate_4dof": 0.0356,
- "ratio": 0.489,
- "rpe_rot_deg": 0.885,
- "nees_yaw": 13.551413172391054,
- "nees_rp": 1.3358872845392693,
- "nees_pos": 2.3877111609347943,
+ "ate_4dof": 0.0225,
+ "ratio": 0.309,
+ "rpe_rot_deg": 0.517,
+ "nees_yaw": 6.51969554474202,
+ "nees_rp": 1.439809470080442,
+ "nees_pos": 0.8557222482517082,
"yaw_sigma_first": 0.1000449929170029,
- "yaw_sigma_min": 0.020892395205771467,
- "yaw_sigma_ratio": 0.2088,
- "yaw_err_max": 0.1436643799388615,
- "nis": 1.00041,
- "ledger_info_yaw": 31715097377.80439,
- "ledger_share_s6": 13598025669.736263
+ "yaw_sigma_min": 0.016654163720140426,
+ "yaw_sigma_ratio": 0.1665,
+ "yaw_err_max": 0.11092917036241866,
+ "nis": 1.0005,
+ "ledger_info_yaw": 16328273906.96167,
+ "ledger_share_s6": 19225618467.116405
},
"V2_02_medium/0.5/2/default": {
"status": "scored",
"ate_4dof": 0.0472,
"ratio": 0.648,
- "rpe_rot_deg": 1.319,
- "nees_yaw": 0.18085881193589035,
- "nees_rp": 1.0392215321321248,
- "nees_pos": 0.23866379006127705,
+ "rpe_rot_deg": 1.008,
+ "nees_yaw": 0.5268795591364305,
+ "nees_rp": 3.0487897906377337,
+ "nees_pos": 0.6230919029307216,
"yaw_sigma_first": 0.1000449929170029,
- "yaw_sigma_min": 0.06408515323660147,
- "yaw_sigma_ratio": 0.6406,
- "yaw_err_max": 0.07787784819924741,
- "nis": 0.0121123,
- "ledger_info_yaw": 360900674.8746695,
- "ledger_share_s6": 170783992.41702172
+ "yaw_sigma_min": 0.04349732693519672,
+ "yaw_sigma_ratio": 0.4348,
+ "yaw_err_max": 0.06099816841568509,
+ "nis": 0.0121912,
+ "ledger_info_yaw": 271595059.77452374,
+ "ledger_share_s6": 295943458.44539654
},
"V2_02_medium/0.5/3/matched": {
"status": "scored",
- "ate_4dof": 0.0279,
- "ratio": 0.383,
- "rpe_rot_deg": 0.739,
- "nees_yaw": 20.375052795765157,
- "nees_rp": 1.9234946494111493,
- "nees_pos": 3.432762247519349,
- "yaw_sigma_first": 0.1000449929169597,
- "yaw_sigma_min": 0.020637610040667245,
- "yaw_sigma_ratio": 0.2063,
- "yaw_err_max": 0.13435969561386513,
- "nis": 1.00255,
- "ledger_info_yaw": 3554315484406.3486,
- "ledger_share_s6": 1936858446221.3315
+ "ate_4dof": 0.0149,
+ "ratio": 0.205,
+ "rpe_rot_deg": 0.384,
+ "nees_yaw": 6.7626777193321015,
+ "nees_rp": 1.9957472163178396,
+ "nees_pos": 0.6897844519088774,
+ "yaw_sigma_first": 0.10004499291695969,
+ "yaw_sigma_min": 0.01460876222651497,
+ "yaw_sigma_ratio": 0.146,
+ "yaw_err_max": 0.09015768309197487,
+ "nis": 1.00283,
+ "ledger_info_yaw": 1565840751364.4175,
+ "ledger_share_s6": 1409759552095.3342
},
"V2_02_medium/0.5/3/default": {
"status": "scored",
- "ate_4dof": 0.0428,
- "ratio": 0.588,
- "rpe_rot_deg": 1.25,
- "nees_yaw": 0.3753988609605499,
- "nees_rp": 1.7255210705247481,
- "nees_pos": 0.2682284426254955,
- "yaw_sigma_first": 0.1000449929169597,
- "yaw_sigma_min": 0.06343307669242072,
- "yaw_sigma_ratio": 0.634,
- "yaw_err_max": 0.0614390846540792,
- "nis": 0.0121558,
- "ledger_info_yaw": 308241479829.8294,
- "ledger_share_s6": 150713443132.43237
+ "ate_4dof": 0.0342,
+ "ratio": 0.47,
+ "rpe_rot_deg": 0.584,
+ "nees_yaw": 2.0584579192020884,
+ "nees_rp": 2.336470348351591,
+ "nees_pos": 0.7835749853006552,
+ "yaw_sigma_first": 0.10004499291695969,
+ "yaw_sigma_min": 0.039823085877173175,
+ "yaw_sigma_ratio": 0.3981,
+ "yaw_err_max": 0.13719371843741815,
+ "nis": 0.0122272,
+ "ledger_info_yaw": 2116801617.0042539,
+ "ledger_share_s6": 2113479916.4238827
},
"V2_03_difficult/1/1/matched": {
"status": "scored",
- "ate_4dof": 0.0453,
- "ratio": 0.251,
- "rpe_rot_deg": 0.203,
- "nees_yaw": 12.168481047821373,
- "nees_rp": 1.9510952738153862,
- "nees_pos": 1.0236233927172764,
- "yaw_sigma_first": 0.10002000052174825,
- "yaw_sigma_min": 0.011147040499610518,
- "yaw_sigma_ratio": 0.1114,
- "yaw_err_max": 0.07128717347285972,
- "nis": 1.01244,
- "ledger_info_yaw": 961673813.5938051,
- "ledger_share_s6": 961673813.5917858
+ "ate_4dof": 0.0213,
+ "ratio": 0.118,
+ "rpe_rot_deg": 0.393,
+ "nees_yaw": 0.13412076805559678,
+ "nees_rp": 2.1197508746938807,
+ "nees_pos": 0.16482079755758008,
+ "yaw_sigma_first": 0.1000200005217483,
+ "yaw_sigma_min": 0.02053596196700854,
+ "yaw_sigma_ratio": 0.2053,
+ "yaw_err_max": 0.03111168622326616,
+ "nis": 1.00284,
+ "ledger_info_yaw": 3133536015.761067,
+ "ledger_share_s6": 3133536015.762779
},
"V2_03_difficult/1/1/default": {
"status": "scored",
- "ate_4dof": 0.0311,
- "ratio": 0.172,
- "rpe_rot_deg": 0.287,
- "nees_yaw": 2.3724907284362056,
- "nees_rp": 1.560252414968542,
- "nees_pos": 0.7313913998864862,
- "yaw_sigma_first": 0.10002000052174825,
- "yaw_sigma_min": 0.028896982789861542,
- "yaw_sigma_ratio": 0.2889,
- "yaw_err_max": 0.057210956430016766,
- "nis": 0.0479076,
- "ledger_info_yaw": 205071338.70767987,
- "ledger_share_s6": 205071338.7079856
+ "ate_4dof": 0.0332,
+ "ratio": 0.184,
+ "rpe_rot_deg": 0.519,
+ "nees_yaw": 13.19379723288357,
+ "nees_rp": 2.1551532374662896,
+ "nees_pos": 2.7218988635827177,
+ "yaw_sigma_first": 0.1000200005217483,
+ "yaw_sigma_min": 0.02375774246100134,
+ "yaw_sigma_ratio": 0.2375,
+ "yaw_err_max": 0.15872751437046656,
+ "nis": 0.0485618,
+ "ledger_info_yaw": 3120318.493277471,
+ "ledger_share_s6": 3120318.493250476
},
"V2_03_difficult/1/2/matched": {
"status": "scored",
- "ate_4dof": 0.5783,
- "ratio": 3.202,
- "rpe_rot_deg": 0.29,
- "nees_yaw": 361.9913363438469,
- "nees_rp": 3.58372820890083,
- "nees_pos": 8.152167764154282,
- "yaw_sigma_first": 0.10002000052174795,
- "yaw_sigma_min": 0.0038986551171578407,
- "yaw_sigma_ratio": 0.039,
- "yaw_err_max": 0.12586365477510392,
- "nis": 1.10848,
- "ledger_info_yaw": 2125799885.8436513,
- "ledger_share_s6": 2125799885.8262124
+ "ate_4dof": 0.0257,
+ "ratio": 0.142,
+ "rpe_rot_deg": 0.248,
+ "nees_yaw": 17.33365886543315,
+ "nees_rp": 2.085655328346726,
+ "nees_pos": 0.9607866620679885,
+ "yaw_sigma_first": 0.100020000521748,
+ "yaw_sigma_min": 0.010021004877458408,
+ "yaw_sigma_ratio": 0.1002,
+ "yaw_err_max": 0.07784014661807576,
+ "nis": 1.00103,
+ "ledger_info_yaw": 6952210987.817904,
+ "ledger_share_s6": 6952210987.807112
},
"V2_03_difficult/1/2/default": {
"status": "scored",
- "ate_4dof": 0.0356,
- "ratio": 0.197,
- "rpe_rot_deg": 0.229,
- "nees_yaw": 0.5809105105443714,
- "nees_rp": 0.8974647928776831,
- "nees_pos": 0.22600592228895738,
- "yaw_sigma_first": 0.10002000052174795,
- "yaw_sigma_min": 0.03023073353171144,
- "yaw_sigma_ratio": 0.3022,
- "yaw_err_max": 0.04029216977121101,
- "nis": 0.0478229,
- "ledger_info_yaw": 102019603.03894909,
- "ledger_share_s6": 102019603.0393468
+ "ate_4dof": 0.0376,
+ "ratio": 0.208,
+ "rpe_rot_deg": 0.574,
+ "nees_yaw": 4.858196322695428,
+ "nees_rp": 2.0607660425962835,
+ "nees_pos": 1.4519624080824778,
+ "yaw_sigma_first": 0.100020000521748,
+ "yaw_sigma_min": 0.025783002434055787,
+ "yaw_sigma_ratio": 0.2578,
+ "yaw_err_max": 0.09820722913629908,
+ "nis": 0.0480233,
+ "ledger_info_yaw": 12518787.265817974,
+ "ledger_share_s6": 12518787.265857175
},
"V2_03_difficult/1/3/matched": {
"status": "scored",
- "ate_4dof": 0.0821,
- "ratio": 0.454,
- "rpe_rot_deg": 0.326,
- "nees_yaw": 9.211970070141096,
- "nees_rp": 2.0238707353713514,
- "nees_pos": 1.4059827674844116,
- "yaw_sigma_first": 0.10002000052173861,
- "yaw_sigma_min": 0.01504447435489236,
- "yaw_sigma_ratio": 0.1504,
- "yaw_err_max": 0.09194978825795763,
- "nis": 1.00449,
- "ledger_info_yaw": 546934463.133219,
- "ledger_share_s6": 546934463.1309006
+ "ate_4dof": 0.0222,
+ "ratio": 0.123,
+ "rpe_rot_deg": 0.412,
+ "nees_yaw": 103.85142825942464,
+ "nees_rp": 2.4123770430948714,
+ "nees_pos": 5.749887270991723,
+ "yaw_sigma_first": 0.10002000052173864,
+ "yaw_sigma_min": 0.0111359803228214,
+ "yaw_sigma_ratio": 0.1113,
+ "yaw_err_max": 0.14395794538654622,
+ "nis": 1.00362,
+ "ledger_info_yaw": 1068188136.4480771,
+ "ledger_share_s6": 1068188136.441402
},
"V2_03_difficult/1/3/default": {
"status": "scored",
- "ate_4dof": 0.0275,
- "ratio": 0.153,
- "rpe_rot_deg": 0.582,
- "nees_yaw": 0.23368302044180955,
- "nees_rp": 1.4443311389052662,
- "nees_pos": 0.20349631775234792,
- "yaw_sigma_first": 0.10002000052173861,
- "yaw_sigma_min": 0.056081675234556976,
- "yaw_sigma_ratio": 0.5607,
- "yaw_err_max": 0.03744186224904956,
- "nis": 0.0479953,
- "ledger_info_yaw": 91376297.6388369,
- "ledger_share_s6": 91376297.63837159
+ "ate_4dof": 0.0308,
+ "ratio": 0.171,
+ "rpe_rot_deg": 0.746,
+ "nees_yaw": 1.710517852790808,
+ "nees_rp": 2.0068509458795183,
+ "nees_pos": 0.6197935885221565,
+ "yaw_sigma_first": 0.10002000052173864,
+ "yaw_sigma_min": 0.024369869817321737,
+ "yaw_sigma_ratio": 0.2436,
+ "yaw_err_max": 0.09184025191310119,
+ "nis": 0.0480851,
+ "ledger_info_yaw": 1203317646030.5188,
+ "ledger_share_s6": 1203317646028.3723
},
"V2_03_difficult/0.5/1/matched": {
"status": "scored",
- "ate_4dof": 0.0187,
- "ratio": 0.104,
- "rpe_rot_deg": 0.445,
- "nees_yaw": 3.95696705798949,
- "nees_rp": 2.0127181560984497,
- "nees_pos": 0.7498641361661444,
- "yaw_sigma_first": 0.10002000052174825,
- "yaw_sigma_min": 0.02092466185666167,
- "yaw_sigma_ratio": 0.2092,
- "yaw_err_max": 0.07859782622653057,
+ "ate_4dof": 0.014,
+ "ratio": 0.077,
+ "rpe_rot_deg": 0.231,
+ "nees_yaw": 11.093132749200759,
+ "nees_rp": 2.036413055542275,
+ "nees_pos": 0.8587405872762465,
+ "yaw_sigma_first": 0.1000200005217483,
+ "yaw_sigma_min": 0.013742858690967242,
+ "yaw_sigma_ratio": 0.1374,
+ "yaw_err_max": 0.09259702843398214,
"nis": 1.00182,
- "ledger_info_yaw": 1016734191.1694574,
- "ledger_share_s6": -694124520.6051099
+ "ledger_info_yaw": 110414203.27592789,
+ "ledger_share_s6": 29486990.031217113
},
"V2_03_difficult/0.5/1/default": {
"status": "scored",
- "ate_4dof": 0.0393,
- "ratio": 0.217,
- "rpe_rot_deg": 1.054,
- "nees_yaw": 0.4738727740984897,
- "nees_rp": 1.432907460917949,
- "nees_pos": 0.27662901895081143,
- "yaw_sigma_first": 0.10002000052174825,
- "yaw_sigma_min": 0.07119432058078139,
- "yaw_sigma_ratio": 0.7118,
- "yaw_err_max": 0.07933823024056758,
- "nis": 0.0120967,
- "ledger_info_yaw": 26200231.460722167,
- "ledger_share_s6": -10380005.541696738
+ "ate_4dof": 0.0259,
+ "ratio": 0.144,
+ "rpe_rot_deg": 0.597,
+ "nees_yaw": 0.4649394223736273,
+ "nees_rp": 2.125288941501017,
+ "nees_pos": 0.31581928131854126,
+ "yaw_sigma_first": 0.1000200005217483,
+ "yaw_sigma_min": 0.03454649665895203,
+ "yaw_sigma_ratio": 0.3454,
+ "yaw_err_max": 0.10449532259219,
+ "nis": 0.0121815,
+ "ledger_info_yaw": 46651330.69693164,
+ "ledger_share_s6": 90471786.60594988
},
"V2_03_difficult/0.5/2/matched": {
"status": "scored",
- "ate_4dof": 0.0212,
- "ratio": 0.118,
- "rpe_rot_deg": 0.633,
- "nees_yaw": 7.913419283506966,
- "nees_rp": 1.5349473703805703,
- "nees_pos": 1.6800312144670755,
- "yaw_sigma_first": 0.10002000052174795,
- "yaw_sigma_min": 0.023376295566819288,
- "yaw_sigma_ratio": 0.2337,
- "yaw_err_max": 0.09680234477458032,
- "nis": 1.00012,
- "ledger_info_yaw": 3018941021.95973,
- "ledger_share_s6": -627433248.2027066
+ "ate_4dof": 0.0159,
+ "ratio": 0.088,
+ "rpe_rot_deg": 0.382,
+ "nees_yaw": 15.195329147958917,
+ "nees_rp": 1.4607345624529737,
+ "nees_pos": 1.914166651310453,
+ "yaw_sigma_first": 0.100020000521748,
+ "yaw_sigma_min": 0.01747980002396516,
+ "yaw_sigma_ratio": 0.1748,
+ "yaw_err_max": 0.08517347089339423,
+ "nis": 1.00023,
+ "ledger_info_yaw": 1078094444.3874853,
+ "ledger_share_s6": 147004181.38695657
},
"V2_03_difficult/0.5/2/default": {
"status": "scored",
- "ate_4dof": 0.0297,
- "ratio": 0.164,
- "rpe_rot_deg": 0.515,
- "nees_yaw": 0.0692051505044159,
- "nees_rp": 0.9899826752455747,
- "nees_pos": 0.08482119489891543,
- "yaw_sigma_first": 0.10002000052174795,
- "yaw_sigma_min": 0.07003235529953879,
- "yaw_sigma_ratio": 0.7002,
- "yaw_err_max": 0.04140844616831907,
- "nis": 0.0121303,
- "ledger_info_yaw": 30490886.06248689,
- "ledger_share_s6": -1574280.3811323883
+ "ate_4dof": 0.0267,
+ "ratio": 0.148,
+ "rpe_rot_deg": 0.63,
+ "nees_yaw": 0.3532995791393874,
+ "nees_rp": 1.8019933216278863,
+ "nees_pos": 0.18915341912147782,
+ "yaw_sigma_first": 0.100020000521748,
+ "yaw_sigma_min": 0.04220067365298826,
+ "yaw_sigma_ratio": 0.4219,
+ "yaw_err_max": 0.0962138280895197,
+ "nis": 0.0121814,
+ "ledger_info_yaw": 4716182.547385052,
+ "ledger_share_s6": 9257386.483885705
},
"V2_03_difficult/0.5/3/matched": {
"status": "scored",
- "ate_4dof": 0.0199,
- "ratio": 0.11,
- "rpe_rot_deg": 0.657,
- "nees_yaw": 21.09619628753277,
- "nees_rp": 1.8780977664887442,
- "nees_pos": 3.6275873167231074,
- "yaw_sigma_first": 0.10002000052173861,
- "yaw_sigma_min": 0.021249833599403755,
- "yaw_sigma_ratio": 0.2125,
- "yaw_err_max": 0.13517929205768475,
- "nis": 1.00327,
- "ledger_info_yaw": 1590656503.8557081,
- "ledger_share_s6": 164268042.11056012
+ "ate_4dof": 0.013,
+ "ratio": 0.072,
+ "rpe_rot_deg": 0.387,
+ "nees_yaw": 84.70490279889417,
+ "nees_rp": 1.9275978682809023,
+ "nees_pos": 7.388409171030284,
+ "yaw_sigma_first": 0.10002000052173864,
+ "yaw_sigma_min": 0.014748327152633672,
+ "yaw_sigma_ratio": 0.1475,
+ "yaw_err_max": 0.17943009165906815,
+ "nis": 1.00349,
+ "ledger_info_yaw": 1144047725.7686856,
+ "ledger_share_s6": 387065342.9434974
},
"V2_03_difficult/0.5/3/default": {
"status": "scored",
- "ate_4dof": 0.0276,
- "ratio": 0.153,
- "rpe_rot_deg": 0.637,
- "nees_yaw": 0.1609778722560655,
- "nees_rp": 1.355950861912466,
- "nees_pos": 0.13386003570074095,
- "yaw_sigma_first": 0.10002000052173861,
- "yaw_sigma_min": 0.06408464199577238,
- "yaw_sigma_ratio": 0.6407,
- "yaw_err_max": 0.05823825726836255,
- "nis": 0.0121847,
- "ledger_info_yaw": 57529123.67856938,
- "ledger_share_s6": 20348664.758472238
+ "ate_4dof": 0.0345,
+ "ratio": 0.191,
+ "rpe_rot_deg": 0.897,
+ "nees_yaw": 4.423108027100867,
+ "nees_rp": 1.6058300127765481,
+ "nees_pos": 0.9712278968302943,
+ "yaw_sigma_first": 0.10002000052173864,
+ "yaw_sigma_min": 0.03240547440444322,
+ "yaw_sigma_ratio": 0.324,
+ "yaw_err_max": 0.11490151795621804,
+ "nis": 0.0122137,
+ "ledger_info_yaw": 7715062.808175538,
+ "ledger_share_s6": 5986430.962732246
}
}
},
"riekf": {
- "commit": "f1551cc",
- "date": "2026-11-09",
+ "commit": "7a38162",
+ "date": "2026-11-10",
"verdict": {
"no_divergence": false,
"matched_beats_default": true,
"no_yaw_collapse": true,
"pass": false,
- "median_yaw_sigma_ratio": 0.9841,
- "median_yaw_nees": 0.011,
+ "median_yaw_sigma_ratio": 0.969,
+ "median_yaw_nees": 0.017,
"median_nis": 1.002,
"truth": "spline (--imu gt, #613)",
"median_ratio": {
"1": {
- "matched": 0.397,
- "default": 0.406
+ "matched": 0.319,
+ "default": 0.534
},
"0.5": {
- "matched": 0.177,
- "default": 0.361
+ "matched": 0.187,
+ "default": 0.399
}
},
"diverged": [
@@ -2945,97 +2945,97 @@
},
"MH_05_difficult/1/1/matched": {
"status": "scored",
- "ate_4dof": 0.041,
- "ratio": 0.09,
- "rpe_rot_deg": 0.166,
- "nees_yaw": 0.001097488273539448,
- "nees_rp": 1.7787975358733934,
- "nees_pos": 0.6584060932617899,
- "yaw_sigma_first": 0.10007997298902055,
- "yaw_sigma_min": 0.10000205754040883,
+ "ate_4dof": 0.0418,
+ "ratio": 0.091,
+ "rpe_rot_deg": 0.167,
+ "nees_yaw": 0.0015892272653912642,
+ "nees_rp": 2.17477401421515,
+ "nees_pos": 0.5862657266625534,
+ "yaw_sigma_first": 0.10007997298902056,
+ "yaw_sigma_min": 0.10000241303537694,
"yaw_sigma_ratio": 0.9992,
- "yaw_err_max": 0.008146520080243563,
- "nis": 1.00183,
+ "yaw_err_max": 0.0074229246635233,
+ "nis": 1.00214,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"MH_05_difficult/1/1/default": {
"status": "scored",
- "ate_4dof": 0.0446,
- "ratio": 0.098,
- "rpe_rot_deg": 0.189,
- "nees_yaw": 0.006411475545263837,
- "nees_rp": 1.3380800306790082,
- "nees_pos": 0.408161028120299,
- "yaw_sigma_first": 0.10007997298902055,
- "yaw_sigma_min": 0.10004103272971783,
- "yaw_sigma_ratio": 0.9996,
- "yaw_err_max": 0.012237935634137838,
- "nis": 0.0477457,
+ "ate_4dof": 0.0468,
+ "ratio": 0.102,
+ "rpe_rot_deg": 0.186,
+ "nees_yaw": 0.0009617725644589735,
+ "nees_rp": 1.5506715664935597,
+ "nees_pos": 0.356060248348215,
+ "yaw_sigma_first": 0.10007997298902056,
+ "yaw_sigma_min": 0.10006424017714033,
+ "yaw_sigma_ratio": 0.9998,
+ "yaw_err_max": 0.00676221186071152,
+ "nis": 0.0477877,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"MH_05_difficult/1/2/matched": {
"status": "scored",
- "ate_4dof": 0.0449,
- "ratio": 0.098,
- "rpe_rot_deg": 0.179,
- "nees_yaw": 0.003886815132094436,
- "nees_rp": 1.0614800681976149,
- "nees_pos": 1.1104711834245655,
- "yaw_sigma_first": 0.10007997298902481,
- "yaw_sigma_min": 0.10000369159502824,
+ "ate_4dof": 0.0456,
+ "ratio": 0.1,
+ "rpe_rot_deg": 0.187,
+ "nees_yaw": 0.009557592579529595,
+ "nees_rp": 1.4654464138869685,
+ "nees_pos": 1.819410101748451,
+ "yaw_sigma_first": 0.1000799729890249,
+ "yaw_sigma_min": 0.10000153927043488,
"yaw_sigma_ratio": 0.9992,
- "yaw_err_max": 0.012002318194400018,
- "nis": 1.00133,
+ "yaw_err_max": 0.016055168614879052,
+ "nis": 1.00135,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"MH_05_difficult/1/2/default": {
"status": "scored",
- "ate_4dof": 0.0931,
- "ratio": 0.204,
- "rpe_rot_deg": 0.353,
- "nees_yaw": 0.00812463814864703,
- "nees_rp": 0.8278581451877249,
- "nees_pos": 1.1284431858275585,
- "yaw_sigma_first": 0.10007997298902481,
- "yaw_sigma_min": 0.10001833421322674,
+ "ate_4dof": 0.0945,
+ "ratio": 0.207,
+ "rpe_rot_deg": 0.33,
+ "nees_yaw": 0.013240179470204493,
+ "nees_rp": 0.8868265357452167,
+ "nees_pos": 0.8774707417062936,
+ "yaw_sigma_first": 0.1000799729890249,
+ "yaw_sigma_min": 0.100016797167589,
"yaw_sigma_ratio": 0.9994,
- "yaw_err_max": 0.016565695398250257,
- "nis": 0.0477449,
+ "yaw_err_max": 0.018532651094185344,
+ "nis": 0.0478087,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"MH_05_difficult/1/3/matched": {
"status": "scored",
- "ate_4dof": 0.0354,
- "ratio": 0.077,
- "rpe_rot_deg": 0.178,
- "nees_yaw": 0.0015795300028258296,
- "nees_rp": 2.104469464846531,
- "nees_pos": 1.0177774355288132,
- "yaw_sigma_first": 0.10007997298900183,
- "yaw_sigma_min": 0.10000067843642012,
+ "ate_4dof": 0.0357,
+ "ratio": 0.078,
+ "rpe_rot_deg": 0.184,
+ "nees_yaw": 0.0006633047140984025,
+ "nees_rp": 2.1120569047020408,
+ "nees_pos": 0.9561720515372116,
+ "yaw_sigma_first": 0.10007997298900187,
+ "yaw_sigma_min": 0.10000275229545612,
"yaw_sigma_ratio": 0.9992,
- "yaw_err_max": 0.00765808530619072,
- "nis": 1.00222,
+ "yaw_err_max": 0.0052094135930788365,
+ "nis": 1.00194,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"MH_05_difficult/1/3/default": {
"status": "scored",
- "ate_4dof": 0.0512,
- "ratio": 0.112,
- "rpe_rot_deg": 0.258,
- "nees_yaw": 0.0006677271912718799,
- "nees_rp": 1.771997367141609,
- "nees_pos": 0.35774899158592055,
- "yaw_sigma_first": 0.10007997298900183,
- "yaw_sigma_min": 0.10007997298900183,
- "yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.005814893116269257,
- "nis": 0.0478268,
+ "ate_4dof": 0.0582,
+ "ratio": 0.127,
+ "rpe_rot_deg": 0.289,
+ "nees_yaw": 0.009597510590288022,
+ "nees_rp": 1.757546769427097,
+ "nees_pos": 0.36154675263861263,
+ "yaw_sigma_first": 0.10007997298900187,
+ "yaw_sigma_min": 0.10006335305605954,
+ "yaw_sigma_ratio": 0.9998,
+ "yaw_err_max": 0.016816249414522352,
+ "nis": 0.0478463,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
@@ -3044,14 +3044,14 @@
"ate_4dof": 0.0185,
"ratio": 0.04,
"rpe_rot_deg": 0.11,
- "nees_yaw": 0.002829235092478655,
- "nees_rp": 2.295330405651249,
- "nees_pos": 0.04039295256924614,
- "yaw_sigma_first": 0.10007997298902055,
- "yaw_sigma_min": 0.09147847061127598,
- "yaw_sigma_ratio": 0.9141,
- "yaw_err_max": 0.02377612974683939,
- "nis": 1.00191,
+ "nees_yaw": 0.016229766207732955,
+ "nees_rp": 2.3183706942016644,
+ "nees_pos": 0.041142848911357194,
+ "yaw_sigma_first": 0.10007997298902056,
+ "yaw_sigma_min": 0.09125350624264127,
+ "yaw_sigma_ratio": 0.9118,
+ "yaw_err_max": 0.03128820708144561,
+ "nis": 1.00192,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
@@ -3060,590 +3060,590 @@
"ate_4dof": 0.0664,
"ratio": 0.145,
"rpe_rot_deg": 0.394,
- "nees_yaw": 0.03519735706065276,
- "nees_rp": 2.0496328057022226,
- "nees_pos": 0.6145151115029858,
- "yaw_sigma_first": 0.10007997298902055,
- "yaw_sigma_min": 0.09143744026085879,
- "yaw_sigma_ratio": 0.9136,
- "yaw_err_max": 0.02977941565965108,
- "nis": 0.0121305,
+ "nees_yaw": 0.05576013745822545,
+ "nees_rp": 2.2605340322462397,
+ "nees_pos": 0.6428033700887632,
+ "yaw_sigma_first": 0.10007997298902056,
+ "yaw_sigma_min": 0.09105306180897413,
+ "yaw_sigma_ratio": 0.9098,
+ "yaw_err_max": 0.03449747677108682,
+ "nis": 0.0121376,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"MH_05_difficult/0.5/2/matched": {
"status": "scored",
- "ate_4dof": 0.0244,
+ "ate_4dof": 0.0243,
"ratio": 0.053,
"rpe_rot_deg": 0.12,
- "nees_yaw": 0.041832439073980446,
- "nees_rp": 1.4199005855258862,
- "nees_pos": 0.052714769719254234,
- "yaw_sigma_first": 0.10007997298902481,
- "yaw_sigma_min": 0.09297817980293846,
- "yaw_sigma_ratio": 0.929,
- "yaw_err_max": 0.027818121642198716,
- "nis": 1.001,
+ "nees_yaw": 0.06117593042447907,
+ "nees_rp": 1.3121482498938273,
+ "nees_pos": 0.05190022197775416,
+ "yaw_sigma_first": 0.1000799729890249,
+ "yaw_sigma_min": 0.09043105581387216,
+ "yaw_sigma_ratio": 0.9036,
+ "yaw_err_max": 0.031423537900751704,
+ "nis": 1.00111,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"MH_05_difficult/0.5/2/default": {
"status": "scored",
- "ate_4dof": 0.1006,
- "ratio": 0.22,
- "rpe_rot_deg": 0.256,
- "nees_yaw": 0.14842901040806203,
- "nees_rp": 0.696601955944146,
- "nees_pos": 0.679719263054174,
- "yaw_sigma_first": 0.10007997298902481,
- "yaw_sigma_min": 0.09424219849315525,
- "yaw_sigma_ratio": 0.9417,
- "yaw_err_max": 0.04627352613057905,
- "nis": 0.0120759,
+ "ate_4dof": 0.1028,
+ "ratio": 0.225,
+ "rpe_rot_deg": 0.272,
+ "nees_yaw": 0.17216754123144185,
+ "nees_rp": 0.84112816430734,
+ "nees_pos": 0.7266351620660337,
+ "yaw_sigma_first": 0.1000799729890249,
+ "yaw_sigma_min": 0.09051508438534761,
+ "yaw_sigma_ratio": 0.9044,
+ "yaw_err_max": 0.04747299440643276,
+ "nis": 0.0121082,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"MH_05_difficult/0.5/3/matched": {
"status": "scored",
- "ate_4dof": 0.0219,
+ "ate_4dof": 0.022,
"ratio": 0.048,
"rpe_rot_deg": 0.105,
- "nees_yaw": 0.015739397515868968,
- "nees_rp": 1.7422587505390348,
- "nees_pos": 0.049359607455637555,
- "yaw_sigma_first": 0.10007997298900183,
- "yaw_sigma_min": 0.0948561734281353,
- "yaw_sigma_ratio": 0.9478,
- "yaw_err_max": 0.016800763307842803,
- "nis": 1.00193,
+ "nees_yaw": 0.005630807190975341,
+ "nees_rp": 1.755055580152385,
+ "nees_pos": 0.05399612099783919,
+ "yaw_sigma_first": 0.10007997298900187,
+ "yaw_sigma_min": 0.09416739154777073,
+ "yaw_sigma_ratio": 0.9409,
+ "yaw_err_max": 0.009580983864776762,
+ "nis": 1.002,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"MH_05_difficult/0.5/3/default": {
"status": "scored",
- "ate_4dof": 0.0589,
- "ratio": 0.129,
- "rpe_rot_deg": 0.175,
- "nees_yaw": 0.019788712467639715,
- "nees_rp": 1.8019224415666542,
- "nees_pos": 0.21788569183768916,
- "yaw_sigma_first": 0.10007997298900183,
- "yaw_sigma_min": 0.09535147527989235,
- "yaw_sigma_ratio": 0.9528,
- "yaw_err_max": 0.019598287977788856,
- "nis": 0.0121104,
+ "ate_4dof": 0.0594,
+ "ratio": 0.13,
+ "rpe_rot_deg": 0.183,
+ "nees_yaw": 0.008162084313694053,
+ "nees_rp": 2.019649657173199,
+ "nees_pos": 0.23750468390720994,
+ "yaw_sigma_first": 0.10007997298900187,
+ "yaw_sigma_min": 0.09471892856986608,
+ "yaw_sigma_ratio": 0.9464,
+ "yaw_err_max": 0.014157199915536378,
+ "nis": 0.0121287,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_01_easy/1/1/matched": {
"status": "scored",
- "ate_4dof": 0.0274,
- "ratio": 0.397,
+ "ate_4dof": 0.0403,
+ "ratio": 0.584,
"rpe_rot_deg": 0.258,
- "nees_yaw": 0.04704803909015082,
- "nees_rp": 1.5683051383128037,
- "nees_pos": 0.31664044959111187,
+ "nees_yaw": 0.035077715142784675,
+ "nees_rp": 2.123117740584596,
+ "nees_pos": 0.7040531307263637,
"yaw_sigma_first": 0.10001125144439284,
- "yaw_sigma_min": 0.1000017041846993,
- "yaw_sigma_ratio": 0.9999,
- "yaw_err_max": 0.027288715106145597,
- "nis": 0.99986,
+ "yaw_sigma_min": 0.10000795555970617,
+ "yaw_sigma_ratio": 1.0,
+ "yaw_err_max": 0.023908214287011693,
+ "nis": 0.999984,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_01_easy/1/1/default": {
"status": "scored",
- "ate_4dof": 0.0491,
- "ratio": 0.711,
- "rpe_rot_deg": 0.335,
- "nees_yaw": 0.043696406215020855,
- "nees_rp": 1.7590777281326144,
- "nees_pos": 0.2546829840850875,
+ "ate_4dof": 0.0508,
+ "ratio": 0.735,
+ "rpe_rot_deg": 0.273,
+ "nees_yaw": 0.01984185564182989,
+ "nees_rp": 2.8621982816819735,
+ "nees_pos": 0.2272036342827462,
"yaw_sigma_first": 0.10001125144439284,
"yaw_sigma_min": 0.10001125144439284,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.03112964561581037,
- "nis": 0.0476948,
+ "yaw_err_max": 0.022438257086881762,
+ "nis": 0.0477599,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_01_easy/1/2/matched": {
"status": "scored",
- "ate_4dof": 0.0323,
- "ratio": 0.467,
- "rpe_rot_deg": 0.229,
- "nees_yaw": 0.016885579676806338,
- "nees_rp": 1.0253244662917058,
- "nees_pos": 0.18770816224231207,
- "yaw_sigma_first": 0.10001125144439334,
- "yaw_sigma_min": 0.10000162772435922,
+ "ate_4dof": 0.0285,
+ "ratio": 0.413,
+ "rpe_rot_deg": 0.239,
+ "nees_yaw": 0.029832411268589553,
+ "nees_rp": 1.3558328332459941,
+ "nees_pos": 0.2892498936415951,
+ "yaw_sigma_first": 0.10001125144439336,
+ "yaw_sigma_min": 0.10000258237966948,
"yaw_sigma_ratio": 0.9999,
- "yaw_err_max": 0.018180822316784646,
- "nis": 1.00065,
+ "yaw_err_max": 0.022089806810675128,
+ "nis": 1.00024,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_01_easy/1/2/default": {
"status": "scored",
- "ate_4dof": 0.0425,
- "ratio": 0.616,
- "rpe_rot_deg": 0.425,
- "nees_yaw": 0.029402815121069723,
- "nees_rp": 0.7239929617913688,
- "nees_pos": 0.21394960294731857,
- "yaw_sigma_first": 0.10001125144439334,
- "yaw_sigma_min": 0.10001125144439334,
+ "ate_4dof": 0.0487,
+ "ratio": 0.705,
+ "rpe_rot_deg": 0.65,
+ "nees_yaw": 0.012149585223233759,
+ "nees_rp": 2.0890022656659126,
+ "nees_pos": 0.2711369117390102,
+ "yaw_sigma_first": 0.10001125144439336,
+ "yaw_sigma_min": 0.10001125144439336,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.025135685196280632,
- "nis": 0.0477298,
+ "yaw_err_max": 0.027203118721241523,
+ "nis": 0.0478727,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_01_easy/1/3/matched": {
"status": "scored",
- "ate_4dof": 0.029,
- "ratio": 0.419,
- "rpe_rot_deg": 0.51,
- "nees_yaw": 0.05107642968964892,
- "nees_rp": 2.3852864055923426,
- "nees_pos": 0.5388905484878749,
- "yaw_sigma_first": 0.10001125144439314,
- "yaw_sigma_min": 0.10000247430509523,
+ "ate_4dof": 0.0343,
+ "ratio": 0.496,
+ "rpe_rot_deg": 0.544,
+ "nees_yaw": 0.054620724544748034,
+ "nees_rp": 2.433987626965995,
+ "nees_pos": 0.4028397145187354,
+ "yaw_sigma_first": 0.10001125144439317,
+ "yaw_sigma_min": 0.10000228354031009,
"yaw_sigma_ratio": 0.9999,
- "yaw_err_max": 0.03488617217015157,
- "nis": 1.00306,
+ "yaw_err_max": 0.035923669656276885,
+ "nis": 1.00499,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_01_easy/1/3/default": {
"status": "scored",
- "ate_4dof": 0.0387,
- "ratio": 0.56,
- "rpe_rot_deg": 0.551,
- "nees_yaw": 0.07554505781398681,
- "nees_rp": 1.6523450791966061,
- "nees_pos": 0.10511710820993515,
- "yaw_sigma_first": 0.10001125144439314,
- "yaw_sigma_min": 0.10001125144439314,
+ "ate_4dof": 0.0402,
+ "ratio": 0.582,
+ "rpe_rot_deg": 0.471,
+ "nees_yaw": 0.053084310936818334,
+ "nees_rp": 2.493374342926325,
+ "nees_pos": 0.5089384749806991,
+ "yaw_sigma_first": 0.10001125144439317,
+ "yaw_sigma_min": 0.10001125144439317,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.039480575376079595,
- "nis": 0.0478592,
+ "yaw_err_max": 0.033546665499316286,
+ "nis": 0.0479714,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_01_easy/0.5/1/matched": {
"status": "scored",
- "ate_4dof": 0.0212,
- "ratio": 0.306,
- "rpe_rot_deg": 0.14,
- "nees_yaw": 0.13603329692712798,
- "nees_rp": 1.5546934195414472,
- "nees_pos": 0.10100446188184409,
+ "ate_4dof": 0.0214,
+ "ratio": 0.309,
+ "rpe_rot_deg": 0.178,
+ "nees_yaw": 0.8474888217722439,
+ "nees_rp": 1.6496061770184816,
+ "nees_pos": 0.06486049560795167,
"yaw_sigma_first": 0.10001125144439284,
- "yaw_sigma_min": 0.0848162809331298,
- "yaw_sigma_ratio": 0.8481,
- "yaw_err_max": 0.03459423025530432,
- "nis": 0.999765,
+ "yaw_sigma_min": 0.08150485458814899,
+ "yaw_sigma_ratio": 0.815,
+ "yaw_err_max": 0.08786765399185338,
+ "nis": 0.999791,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_01_easy/0.5/1/default": {
"status": "scored",
- "ate_4dof": 0.0417,
- "ratio": 0.603,
- "rpe_rot_deg": 0.27,
- "nees_yaw": 0.010320239376615492,
- "nees_rp": 1.860602219097575,
- "nees_pos": 0.15750226861227984,
+ "ate_4dof": 0.0426,
+ "ratio": 0.617,
+ "rpe_rot_deg": 0.324,
+ "nees_yaw": 0.33474479317704264,
+ "nees_rp": 2.8058049275307724,
+ "nees_pos": 0.159323001429404,
"yaw_sigma_first": 0.10001125144439284,
- "yaw_sigma_min": 0.08797662728876662,
- "yaw_sigma_ratio": 0.8797,
- "yaw_err_max": 0.028309443070814542,
- "nis": 0.012069,
+ "yaw_sigma_min": 0.08811434490647668,
+ "yaw_sigma_ratio": 0.881,
+ "yaw_err_max": 0.08786765399185338,
+ "nis": 0.0121144,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_01_easy/0.5/2/matched": {
"status": "scored",
- "ate_4dof": 0.0228,
- "ratio": 0.33,
- "rpe_rot_deg": 0.18,
- "nees_yaw": 0.03676186411964229,
- "nees_rp": 1.0716164120171499,
- "nees_pos": 0.07713352892935099,
- "yaw_sigma_first": 0.10001125144439334,
- "yaw_sigma_min": 0.089870303460673,
- "yaw_sigma_ratio": 0.8986,
- "yaw_err_max": 0.02616786238199398,
- "nis": 1.00061,
+ "ate_4dof": 0.023,
+ "ratio": 0.333,
+ "rpe_rot_deg": 0.217,
+ "nees_yaw": 0.00753823771586045,
+ "nees_rp": 1.6609471963427531,
+ "nees_pos": 0.0923406227881325,
+ "yaw_sigma_first": 0.10001125144439336,
+ "yaw_sigma_min": 0.07757082323509656,
+ "yaw_sigma_ratio": 0.7756,
+ "yaw_err_max": 0.06017725214263296,
+ "nis": 1.00076,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_01_easy/0.5/2/default": {
"status": "scored",
- "ate_4dof": 0.0439,
- "ratio": 0.636,
- "rpe_rot_deg": 0.709,
- "nees_yaw": 0.010304770007794001,
- "nees_rp": 0.805355838155614,
- "nees_pos": 0.1457922524540121,
- "yaw_sigma_first": 0.10001125144439334,
- "yaw_sigma_min": 0.09052365829309787,
- "yaw_sigma_ratio": 0.9051,
- "yaw_err_max": 0.029739454548208893,
- "nis": 0.0120862,
+ "ate_4dof": 0.0438,
+ "ratio": 0.633,
+ "rpe_rot_deg": 0.681,
+ "nees_yaw": 0.011844086845980819,
+ "nees_rp": 1.683977780656431,
+ "nees_pos": 0.1586885842567018,
+ "yaw_sigma_first": 0.10001125144439336,
+ "yaw_sigma_min": 0.08469053672140503,
+ "yaw_sigma_ratio": 0.8468,
+ "yaw_err_max": 0.0695245041437982,
+ "nis": 0.0121694,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_01_easy/0.5/3/matched": {
"status": "scored",
- "ate_4dof": 0.0161,
- "ratio": 0.233,
- "rpe_rot_deg": 0.277,
- "nees_yaw": 0.008591846436734659,
- "nees_rp": 2.281223932731895,
- "nees_pos": 0.08421933127528193,
- "yaw_sigma_first": 0.10001125144439314,
- "yaw_sigma_min": 0.0884078428242202,
- "yaw_sigma_ratio": 0.884,
- "yaw_err_max": 0.024062604846716185,
- "nis": 1.00376,
+ "ate_4dof": 0.0162,
+ "ratio": 0.235,
+ "rpe_rot_deg": 0.294,
+ "nees_yaw": 0.21587389712692762,
+ "nees_rp": 2.3682809777369873,
+ "nees_pos": 0.11778559264702733,
+ "yaw_sigma_first": 0.10001125144439317,
+ "yaw_sigma_min": 0.08181010121261102,
+ "yaw_sigma_ratio": 0.818,
+ "yaw_err_max": 0.05648061081119979,
+ "nis": 1.00387,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_01_easy/0.5/3/default": {
"status": "scored",
- "ate_4dof": 0.0373,
- "ratio": 0.539,
- "rpe_rot_deg": 0.325,
- "nees_yaw": 0.008286514674377675,
- "nees_rp": 1.8696161187220548,
- "nees_pos": 0.2176168890511957,
- "yaw_sigma_first": 0.10001125144439314,
- "yaw_sigma_min": 0.09013503530685257,
- "yaw_sigma_ratio": 0.9012,
- "yaw_err_max": 0.03004356998514906,
- "nis": 0.0121212,
+ "ate_4dof": 0.0368,
+ "ratio": 0.533,
+ "rpe_rot_deg": 0.275,
+ "nees_yaw": 0.08400857175509273,
+ "nees_rp": 2.3530918851809792,
+ "nees_pos": 0.20490188683080054,
+ "yaw_sigma_first": 0.10001125144439317,
+ "yaw_sigma_min": 0.08752373231511802,
+ "yaw_sigma_ratio": 0.8751,
+ "yaw_err_max": 0.05648061081119979,
+ "nis": 0.012173,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_02_medium/1/1/matched": {
"status": "scored",
- "ate_4dof": 0.0303,
- "ratio": 0.519,
- "rpe_rot_deg": 0.27,
- "nees_yaw": 0.0201389908017589,
- "nees_rp": 2.1803620884453747,
- "nees_pos": 1.3416312067501002,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.1000010561253219,
+ "ate_4dof": 0.0271,
+ "ratio": 0.466,
+ "rpe_rot_deg": 0.251,
+ "nees_yaw": 0.017431440137804392,
+ "nees_rp": 2.1602258499957414,
+ "nees_pos": 0.3010821755811967,
+ "yaw_sigma_first": 0.10000125068941465,
+ "yaw_sigma_min": 0.10000093886181494,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.021356743833518592,
- "nis": 1.00341,
+ "yaw_err_max": 0.019990033557688906,
+ "nis": 1.00307,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_02_medium/1/1/default": {
"status": "scored",
- "ate_4dof": 0.0237,
- "ratio": 0.406,
- "rpe_rot_deg": 0.189,
- "nees_yaw": 0.021180783501752024,
- "nees_rp": 1.8494693849725752,
- "nees_pos": 0.12323375110350414,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.10000125068941464,
+ "ate_4dof": 0.027,
+ "ratio": 0.463,
+ "rpe_rot_deg": 0.258,
+ "nees_yaw": 0.035913728475228494,
+ "nees_rp": 2.367749524364115,
+ "nees_pos": 0.445117209002127,
+ "yaw_sigma_first": 0.10000125068941465,
+ "yaw_sigma_min": 0.10000125068941465,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.020367467600583448,
- "nis": 0.0478231,
+ "yaw_err_max": 0.025416934707006823,
+ "nis": 0.0478591,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_02_medium/1/2/matched": {
"status": "scored",
- "ate_4dof": 0.0858,
- "ratio": 1.471,
- "rpe_rot_deg": 0.251,
- "nees_yaw": 0.021737737534629323,
- "nees_rp": 1.9064394941450808,
- "nees_pos": 0.6172324105359502,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.10000125068941464,
+ "ate_4dof": 0.0266,
+ "ratio": 0.455,
+ "rpe_rot_deg": 0.257,
+ "nees_yaw": 0.02632849133754459,
+ "nees_rp": 2.2754644470396155,
+ "nees_pos": 0.1999528013242281,
+ "yaw_sigma_first": 0.10000125068941466,
+ "yaw_sigma_min": 0.10000125068941466,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.02165453569742021,
- "nis": 1.00203,
+ "yaw_err_max": 0.022987042166015572,
+ "nis": 0.999184,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_02_medium/1/2/default": {
"status": "scored",
- "ate_4dof": 0.0215,
- "ratio": 0.369,
- "rpe_rot_deg": 0.255,
- "nees_yaw": 0.03198028782664397,
- "nees_rp": 1.221935326673243,
- "nees_pos": 0.09194324477127652,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.10000125068941464,
+ "ate_4dof": 0.0364,
+ "ratio": 0.624,
+ "rpe_rot_deg": 0.329,
+ "nees_yaw": 0.04331946984569129,
+ "nees_rp": 2.3885232385453032,
+ "nees_pos": 0.20321219794381415,
+ "yaw_sigma_first": 0.10000125068941466,
+ "yaw_sigma_min": 0.10000125068941466,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.024955744176609212,
- "nis": 0.0476674,
+ "yaw_err_max": 0.028702273943040396,
+ "nis": 0.0477775,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_02_medium/1/3/matched": {
"status": "scored",
- "ate_4dof": 0.3565,
- "ratio": 6.116,
- "rpe_rot_deg": 0.227,
- "nees_yaw": 0.02850010584001307,
- "nees_rp": 1.6497114232170338,
- "nees_pos": 1.3707802121794335,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.10000062081686126,
+ "ate_4dof": 0.025,
+ "ratio": 0.429,
+ "rpe_rot_deg": 0.197,
+ "nees_yaw": 0.02461771160291855,
+ "nees_rp": 1.616495107029593,
+ "nees_pos": 0.12827682545412572,
+ "yaw_sigma_first": 0.10000125068941462,
+ "yaw_sigma_min": 0.10000090127508773,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.02791657657947247,
- "nis": 1.02021,
+ "yaw_err_max": 0.02666918135494474,
+ "nis": 1.00543,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_02_medium/1/3/default": {
"status": "scored",
- "ate_4dof": 0.0247,
- "ratio": 0.424,
- "rpe_rot_deg": 0.241,
- "nees_yaw": 0.022370913271919375,
- "nees_rp": 0.7074309922803347,
- "nees_pos": 0.13053346144196237,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.10000125068941464,
+ "ate_4dof": 0.0334,
+ "ratio": 0.572,
+ "rpe_rot_deg": 0.32,
+ "nees_yaw": 0.05357966802715537,
+ "nees_rp": 1.374505064634366,
+ "nees_pos": 0.26020876781168534,
+ "yaw_sigma_first": 0.10000125068941462,
+ "yaw_sigma_min": 0.10000125068941462,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.025435147035141027,
- "nis": 0.0479499,
+ "yaw_err_max": 0.03474541432977702,
+ "nis": 0.0480146,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_02_medium/0.5/1/matched": {
"status": "scored",
- "ate_4dof": 0.0211,
- "ratio": 0.362,
- "rpe_rot_deg": 0.156,
- "nees_yaw": 0.05734009766125271,
- "nees_rp": 2.398469906258603,
- "nees_pos": 0.11417443434894932,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.09410046016547426,
- "yaw_sigma_ratio": 0.941,
- "yaw_err_max": 0.04413473904711352,
- "nis": 1.00319,
+ "ate_4dof": 0.0213,
+ "ratio": 0.366,
+ "rpe_rot_deg": 0.175,
+ "nees_yaw": 0.12277434322545162,
+ "nees_rp": 2.5302067209936636,
+ "nees_pos": 0.1176316116843327,
+ "yaw_sigma_first": 0.10000125068941465,
+ "yaw_sigma_min": 0.09049044258976467,
+ "yaw_sigma_ratio": 0.9049,
+ "yaw_err_max": 0.03624268754277917,
+ "nis": 1.00332,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_02_medium/0.5/1/default": {
"status": "scored",
- "ate_4dof": 0.023,
- "ratio": 0.394,
- "rpe_rot_deg": 0.184,
- "nees_yaw": 0.06055139928242892,
- "nees_rp": 2.61535427693816,
- "nees_pos": 0.12513591370584679,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.09534692054253743,
- "yaw_sigma_ratio": 0.9535,
- "yaw_err_max": 0.03052952141347985,
- "nis": 0.0121256,
+ "ate_4dof": 0.0248,
+ "ratio": 0.425,
+ "rpe_rot_deg": 0.256,
+ "nees_yaw": 0.1551487378149693,
+ "nees_rp": 2.826822349065177,
+ "nees_pos": 0.12923707893212333,
+ "yaw_sigma_first": 0.10000125068941465,
+ "yaw_sigma_min": 0.09333707879909678,
+ "yaw_sigma_ratio": 0.9334,
+ "yaw_err_max": 0.04276662565178319,
+ "nis": 0.0121365,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_02_medium/0.5/2/matched": {
"status": "scored",
- "ate_4dof": 0.0137,
- "ratio": 0.235,
- "rpe_rot_deg": 0.14,
- "nees_yaw": 0.02747562147039328,
- "nees_rp": 1.4674349485717582,
- "nees_pos": 0.07923740180207003,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.09420851082825189,
- "yaw_sigma_ratio": 0.9421,
- "yaw_err_max": 0.021581511783272157,
- "nis": 0.999472,
+ "ate_4dof": 0.0134,
+ "ratio": 0.229,
+ "rpe_rot_deg": 0.15,
+ "nees_yaw": 0.03286068864086208,
+ "nees_rp": 1.4407288605711395,
+ "nees_pos": 0.07412375317781061,
+ "yaw_sigma_first": 0.10000125068941466,
+ "yaw_sigma_min": 0.08586359497939633,
+ "yaw_sigma_ratio": 0.8586,
+ "yaw_err_max": 0.03103867952945239,
+ "nis": 0.999638,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_02_medium/0.5/2/default": {
"status": "scored",
- "ate_4dof": 0.0168,
- "ratio": 0.288,
- "rpe_rot_deg": 0.177,
- "nees_yaw": 0.0012480002137470057,
- "nees_rp": 1.5580746905795524,
- "nees_pos": 0.028907735419108595,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.09527607813643926,
- "yaw_sigma_ratio": 0.9527,
- "yaw_err_max": 0.021581511783272157,
- "nis": 0.0121066,
+ "ate_4dof": 0.0217,
+ "ratio": 0.372,
+ "rpe_rot_deg": 0.249,
+ "nees_yaw": 0.011353401509212479,
+ "nees_rp": 2.1094214433427156,
+ "nees_pos": 0.052241040179770507,
+ "yaw_sigma_first": 0.10000125068941466,
+ "yaw_sigma_min": 0.08914700641961931,
+ "yaw_sigma_ratio": 0.8915,
+ "yaw_err_max": 0.03103867952945239,
+ "nis": 0.0121855,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_02_medium/0.5/3/matched": {
"status": "scored",
- "ate_4dof": 0.0125,
- "ratio": 0.215,
- "rpe_rot_deg": 0.135,
- "nees_yaw": 0.007684837646938256,
- "nees_rp": 1.8416558614872618,
- "nees_pos": 0.04028109837696418,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.09556932591159617,
- "yaw_sigma_ratio": 0.9557,
- "yaw_err_max": 0.01357753956869808,
- "nis": 1.00619,
+ "ate_4dof": 0.0124,
+ "ratio": 0.212,
+ "rpe_rot_deg": 0.133,
+ "nees_yaw": 0.028669620357312353,
+ "nees_rp": 1.7083995862559727,
+ "nees_pos": 0.04266698640572973,
+ "yaw_sigma_first": 0.10000125068941462,
+ "yaw_sigma_min": 0.09063920772006151,
+ "yaw_sigma_ratio": 0.9064,
+ "yaw_err_max": 0.023482133735445058,
+ "nis": 1.00644,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_02_medium/0.5/3/default": {
"status": "scored",
- "ate_4dof": 0.0212,
- "ratio": 0.364,
- "rpe_rot_deg": 0.152,
- "nees_yaw": 0.01671691501779485,
- "nees_rp": 0.9660091149083863,
- "nees_pos": 0.07115127015669816,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.09627161366913366,
- "yaw_sigma_ratio": 0.9627,
- "yaw_err_max": 0.014522651915205086,
- "nis": 0.0121858,
+ "ate_4dof": 0.0245,
+ "ratio": 0.419,
+ "rpe_rot_deg": 0.218,
+ "nees_yaw": 0.05280428984571363,
+ "nees_rp": 1.3625877796089407,
+ "nees_pos": 0.083842032984126,
+ "yaw_sigma_first": 0.10000125068941462,
+ "yaw_sigma_min": 0.09303614300682002,
+ "yaw_sigma_ratio": 0.9303,
+ "yaw_err_max": 0.025247723139274713,
+ "nis": 0.0122264,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_03_difficult/1/1/matched": {
"status": "scored",
- "ate_4dof": 0.1305,
- "ratio": 1.954,
- "rpe_rot_deg": 0.212,
- "nees_yaw": 0.019253224209847007,
- "nees_rp": 3.2607350017805716,
- "nees_pos": 0.6971228523359728,
+ "ate_4dof": 0.0177,
+ "ratio": 0.265,
+ "rpe_rot_deg": 0.228,
+ "nees_yaw": 0.0024510791054251657,
+ "nees_rp": 3.164805487073741,
+ "nees_pos": 0.7400359253621183,
"yaw_sigma_first": 0.10000125068941464,
"yaw_sigma_min": 0.10000125068941464,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.01762694972484468,
- "nis": 1.00279,
+ "yaw_err_max": 0.008900136188617534,
+ "nis": 1.00232,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_03_difficult/1/1/default": {
"status": "scored",
- "ate_4dof": 0.0262,
- "ratio": 0.393,
- "rpe_rot_deg": 0.242,
- "nees_yaw": 0.011465546882812715,
- "nees_rp": 2.043075874276082,
- "nees_pos": 0.18291096680084107,
+ "ate_4dof": 0.0317,
+ "ratio": 0.474,
+ "rpe_rot_deg": 0.373,
+ "nees_yaw": 0.00560204634511027,
+ "nees_rp": 2.3795136514902886,
+ "nees_pos": 1.1018079455649492,
"yaw_sigma_first": 0.10000125068941464,
"yaw_sigma_min": 0.10000125068941464,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.017197514310119507,
- "nis": 0.0477928,
+ "yaw_err_max": 0.01697570102382515,
+ "nis": 0.0478606,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_03_difficult/1/2/matched": {
"status": "scored",
- "ate_4dof": 0.0542,
- "ratio": 0.812,
- "rpe_rot_deg": 0.242,
- "nees_yaw": 0.007322369259081876,
- "nees_rp": 1.6077443838188412,
- "nees_pos": 0.3239218243936558,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.10000110058051503,
+ "ate_4dof": 0.0277,
+ "ratio": 0.415,
+ "rpe_rot_deg": 0.232,
+ "nees_yaw": 0.01232999462715086,
+ "nees_rp": 1.7429078858773788,
+ "nees_pos": 0.6130796283020301,
+ "yaw_sigma_first": 0.10000125068941466,
+ "yaw_sigma_min": 0.10000120840525976,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.015830026391405433,
- "nis": 0.999297,
+ "yaw_err_max": 0.016806612482285807,
+ "nis": 0.998849,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_03_difficult/1/2/default": {
"status": "scored",
- "ate_4dof": 0.026,
- "ratio": 0.389,
- "rpe_rot_deg": 0.386,
- "nees_yaw": 0.002524244617912509,
- "nees_rp": 0.7484212413270971,
- "nees_pos": 0.14647233857761713,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.10000125068941464,
+ "ate_4dof": 0.0448,
+ "ratio": 0.671,
+ "rpe_rot_deg": 0.479,
+ "nees_yaw": 0.005983547006656635,
+ "nees_rp": 1.2796338580839854,
+ "nees_pos": 2.2987263829610773,
+ "yaw_sigma_first": 0.10000125068941466,
+ "yaw_sigma_min": 0.10000125068941466,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.013721572932148802,
- "nis": 0.0476578,
+ "yaw_err_max": 0.01417199805084897,
+ "nis": 0.0478361,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_03_difficult/1/3/matched": {
"status": "scored",
- "ate_4dof": 0.0865,
- "ratio": 1.295,
- "rpe_rot_deg": 0.195,
- "nees_yaw": 0.010620303495647711,
- "nees_rp": 2.6813216210392326,
- "nees_pos": 0.8354616232868122,
+ "ate_4dof": 0.02,
+ "ratio": 0.299,
+ "rpe_rot_deg": 0.242,
+ "nees_yaw": 0.016750628327705585,
+ "nees_rp": 2.7525772450494377,
+ "nees_pos": 0.7985678799406729,
"yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.10000079137130105,
+ "yaw_sigma_min": 0.10000047244195438,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.016750920578635083,
- "nis": 1.00406,
+ "yaw_err_max": 0.01737497493805628,
+ "nis": 1.00299,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_03_difficult/1/3/default": {
"status": "scored",
- "ate_4dof": 0.0289,
- "ratio": 0.432,
- "rpe_rot_deg": 0.273,
- "nees_yaw": 0.04391536888012641,
- "nees_rp": 1.88134640796426,
- "nees_pos": 0.22675912507996088,
+ "ate_4dof": 0.0379,
+ "ratio": 0.567,
+ "rpe_rot_deg": 0.343,
+ "nees_yaw": 0.020137504804754326,
+ "nees_rp": 2.2315807419538354,
+ "nees_pos": 1.5802557696075206,
"yaw_sigma_first": 0.10000125068941464,
"yaw_sigma_min": 0.10000125068941464,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.03092435642634007,
- "nis": 0.0478295,
+ "yaw_err_max": 0.023155593203559092,
+ "nis": 0.0479136,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_03_difficult/0.5/1/matched": {
"status": "scored",
- "ate_4dof": 0.0115,
- "ratio": 0.172,
- "rpe_rot_deg": 0.157,
- "nees_yaw": 0.057022323210878134,
- "nees_rp": 3.9290676134932894,
- "nees_pos": 0.07623639699935482,
+ "ate_4dof": 0.0109,
+ "ratio": 0.163,
+ "rpe_rot_deg": 0.169,
+ "nees_yaw": 0.8273937723173901,
+ "nees_rp": 4.022679920219149,
+ "nees_pos": 0.07167920262954088,
"yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.09439186207805751,
- "yaw_sigma_ratio": 0.9439,
- "yaw_err_max": 0.034327806300557685,
- "nis": 1.00244,
+ "yaw_sigma_min": 0.09051513424783066,
+ "yaw_sigma_ratio": 0.9051,
+ "yaw_err_max": 0.08699108259328314,
+ "nis": 1.00249,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_03_difficult/0.5/1/default": {
"status": "scored",
- "ate_4dof": 0.0241,
- "ratio": 0.361,
- "rpe_rot_deg": 0.125,
- "nees_yaw": 0.02344923303389398,
- "nees_rp": 2.077221480218381,
- "nees_pos": 0.10928652653119998,
+ "ate_4dof": 0.0259,
+ "ratio": 0.388,
+ "rpe_rot_deg": 0.19,
+ "nees_yaw": 0.3167834808865485,
+ "nees_rp": 2.2803789517152246,
+ "nees_pos": 0.1342265829125643,
"yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.09589957210110814,
- "yaw_sigma_ratio": 0.959,
- "yaw_err_max": 0.021938792612839403,
- "nis": 0.0120763,
+ "yaw_sigma_min": 0.0935112330500839,
+ "yaw_sigma_ratio": 0.9351,
+ "yaw_err_max": 0.05930274214086998,
+ "nis": 0.0120971,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
@@ -3651,639 +3651,639 @@
"status": "scored",
"ate_4dof": 0.0099,
"ratio": 0.148,
- "rpe_rot_deg": 0.129,
- "nees_yaw": 0.00196043750297858,
- "nees_rp": 1.4737355036069935,
- "nees_pos": 0.04269611611811271,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.09394517193611794,
- "yaw_sigma_ratio": 0.9394,
- "yaw_err_max": 0.01966284952880038,
- "nis": 0.998994,
+ "rpe_rot_deg": 0.211,
+ "nees_yaw": 0.01604947360591479,
+ "nees_rp": 1.8744967496134957,
+ "nees_pos": 0.04003912314476687,
+ "yaw_sigma_first": 0.10000125068941466,
+ "yaw_sigma_min": 0.08488131338768022,
+ "yaw_sigma_ratio": 0.8488,
+ "yaw_err_max": 0.042219723675336975,
+ "nis": 0.999163,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_03_difficult/0.5/2/default": {
"status": "scored",
- "ate_4dof": 0.0194,
- "ratio": 0.291,
- "rpe_rot_deg": 0.135,
- "nees_yaw": 0.005769380392857944,
- "nees_rp": 0.7727448082939784,
- "nees_pos": 0.09287603822807217,
- "yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.09497623022077056,
- "yaw_sigma_ratio": 0.9498,
- "yaw_err_max": 0.01966284952880038,
- "nis": 0.0120484,
+ "ate_4dof": 0.0244,
+ "ratio": 0.365,
+ "rpe_rot_deg": 0.151,
+ "nees_yaw": 0.03343931677830183,
+ "nees_rp": 1.1831535326468468,
+ "nees_pos": 0.12304308614877799,
+ "yaw_sigma_first": 0.10000125068941466,
+ "yaw_sigma_min": 0.08884972436849348,
+ "yaw_sigma_ratio": 0.8885,
+ "yaw_err_max": 0.042219723675336975,
+ "nis": 0.012098,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_03_difficult/0.5/3/matched": {
"status": "scored",
- "ate_4dof": 0.0121,
- "ratio": 0.181,
- "rpe_rot_deg": 0.149,
- "nees_yaw": 0.002917272540000732,
- "nees_rp": 2.4497973431907085,
- "nees_pos": 0.02466320712624135,
+ "ate_4dof": 0.0118,
+ "ratio": 0.177,
+ "rpe_rot_deg": 0.126,
+ "nees_yaw": 0.0027059027579864237,
+ "nees_rp": 2.332233157017906,
+ "nees_pos": 0.019711013986162033,
"yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.09540588467239529,
- "yaw_sigma_ratio": 0.954,
- "yaw_err_max": 0.012654805847851005,
- "nis": 1.00348,
+ "yaw_sigma_min": 0.08957526778984942,
+ "yaw_sigma_ratio": 0.8957,
+ "yaw_err_max": 0.029115516993946947,
+ "nis": 1.00351,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V1_03_difficult/0.5/3/default": {
"status": "scored",
- "ate_4dof": 0.026,
- "ratio": 0.39,
- "rpe_rot_deg": 0.184,
- "nees_yaw": 0.013848875111769866,
- "nees_rp": 1.9606083834268593,
- "nees_pos": 0.06285639719630993,
+ "ate_4dof": 0.0289,
+ "ratio": 0.433,
+ "rpe_rot_deg": 0.199,
+ "nees_yaw": 0.007513641649091275,
+ "nees_rp": 2.019449258795012,
+ "nees_pos": 0.06773955706304131,
"yaw_sigma_first": 0.10000125068941464,
- "yaw_sigma_min": 0.09607640260834321,
- "yaw_sigma_ratio": 0.9608,
- "yaw_err_max": 0.01588347165244804,
- "nis": 0.0121134,
+ "yaw_sigma_min": 0.09331282039312701,
+ "yaw_sigma_ratio": 0.9331,
+ "yaw_err_max": 0.029115516993946947,
+ "nis": 0.0121464,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_01_easy/1/1/matched": {
"status": "scored",
- "ate_4dof": 0.0227,
- "ratio": 0.179,
- "rpe_rot_deg": 0.255,
- "nees_yaw": 0.00043029295171598416,
- "nees_rp": 1.63754893929531,
- "nees_pos": 0.33630973728786956,
- "yaw_sigma_first": 0.10003124796018417,
- "yaw_sigma_min": 0.10000240760662861,
+ "ate_4dof": 0.0208,
+ "ratio": 0.164,
+ "rpe_rot_deg": 0.445,
+ "nees_yaw": 0.0016508741137493941,
+ "nees_rp": 2.513483263310325,
+ "nees_pos": 0.23767713061846774,
+ "yaw_sigma_first": 0.1000312479601842,
+ "yaw_sigma_min": 0.10000167478848665,
"yaw_sigma_ratio": 0.9997,
- "yaw_err_max": 0.0043845214014332865,
- "nis": 1.00201,
+ "yaw_err_max": 0.007976082970924822,
+ "nis": 1.0043,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_01_easy/1/1/default": {
"status": "scored",
- "ate_4dof": 0.0216,
- "ratio": 0.17,
- "rpe_rot_deg": 0.261,
- "nees_yaw": 0.0005784074238108268,
- "nees_rp": 1.6577411766743828,
- "nees_pos": 0.09480931178983064,
- "yaw_sigma_first": 0.10003124796018417,
- "yaw_sigma_min": 0.10003124796018417,
+ "ate_4dof": 0.0292,
+ "ratio": 0.23,
+ "rpe_rot_deg": 0.465,
+ "nees_yaw": 0.002893321505120911,
+ "nees_rp": 4.545255505389554,
+ "nees_pos": 3.357572231180365,
+ "yaw_sigma_first": 0.1000312479601842,
+ "yaw_sigma_min": 0.1000312479601842,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.005390313937105901,
- "nis": 0.0477917,
+ "yaw_err_max": 0.010450336763130282,
+ "nis": 0.0480166,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_01_easy/1/2/matched": {
"status": "scored",
- "ate_4dof": 0.0376,
- "ratio": 0.297,
- "rpe_rot_deg": 0.895,
- "nees_yaw": 0.011006339861627728,
- "nees_rp": 6.7525114965287765,
- "nees_pos": 1.1425728011097676,
- "yaw_sigma_first": 0.10003124796017195,
- "yaw_sigma_min": 0.10000809323330613,
- "yaw_sigma_ratio": 0.9998,
- "yaw_err_max": 0.014945878957967203,
- "nis": 1.00247,
+ "ate_4dof": 0.0314,
+ "ratio": 0.248,
+ "rpe_rot_deg": 0.712,
+ "nees_yaw": 0.010470917048077892,
+ "nees_rp": 2.480368623837182,
+ "nees_pos": 0.2519067359101842,
+ "yaw_sigma_first": 0.10003124796017199,
+ "yaw_sigma_min": 0.10000142075523134,
+ "yaw_sigma_ratio": 0.9997,
+ "yaw_err_max": 0.01467030254156207,
+ "nis": 1.00306,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_01_easy/1/2/default": {
"status": "scored",
- "ate_4dof": 0.0386,
- "ratio": 0.305,
- "rpe_rot_deg": 0.232,
- "nees_yaw": 0.007164231998610198,
- "nees_rp": 1.0363942612200758,
- "nees_pos": 0.3236934671492645,
- "yaw_sigma_first": 0.10003124796017195,
- "yaw_sigma_min": 0.10003124796017195,
- "yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.015071386893082253,
- "nis": 0.047711,
+ "ate_4dof": 0.0391,
+ "ratio": 0.308,
+ "rpe_rot_deg": 0.626,
+ "nees_yaw": 0.004392532296100314,
+ "nees_rp": 2.135663476136223,
+ "nees_pos": 0.4873953887883146,
+ "yaw_sigma_first": 0.10003124796017199,
+ "yaw_sigma_min": 0.10002350211074393,
+ "yaw_sigma_ratio": 0.9999,
+ "yaw_err_max": 0.014374097187307228,
+ "nis": 0.0481398,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_01_easy/1/3/matched": {
"status": "scored",
- "ate_4dof": 0.0351,
- "ratio": 0.277,
- "rpe_rot_deg": 0.346,
- "nees_yaw": 0.0012848858306213178,
- "nees_rp": 2.022871946275929,
- "nees_pos": 0.8470041829709043,
- "yaw_sigma_first": 0.10003124796016664,
- "yaw_sigma_min": 0.10000108428001689,
+ "ate_4dof": 0.0362,
+ "ratio": 0.285,
+ "rpe_rot_deg": 0.653,
+ "nees_yaw": 0.0017387188941727734,
+ "nees_rp": 2.5077590014032767,
+ "nees_pos": 0.31632856792950026,
+ "yaw_sigma_first": 0.10003124796016667,
+ "yaw_sigma_min": 0.10000337991278789,
"yaw_sigma_ratio": 0.9997,
- "yaw_err_max": 0.026730873368313285,
- "nis": 1.00579,
+ "yaw_err_max": 0.008797532799414072,
+ "nis": 1.00677,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_01_easy/1/3/default": {
"status": "scored",
- "ate_4dof": 0.0449,
- "ratio": 0.354,
- "rpe_rot_deg": 0.493,
- "nees_yaw": 0.006842350130711123,
- "nees_rp": 2.123148412070125,
- "nees_pos": 0.28259901219008693,
- "yaw_sigma_first": 0.10003124796016664,
- "yaw_sigma_min": 0.10003124796016664,
+ "ate_4dof": 0.0474,
+ "ratio": 0.374,
+ "rpe_rot_deg": 0.676,
+ "nees_yaw": 0.004512829673795666,
+ "nees_rp": 2.6094808092558326,
+ "nees_pos": 2.7414014022283317,
+ "yaw_sigma_first": 0.10003124796016667,
+ "yaw_sigma_min": 0.10003124796016667,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.016596606758457883,
- "nis": 0.0479402,
+ "yaw_err_max": 0.016535463536555516,
+ "nis": 0.048238,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_01_easy/0.5/1/matched": {
"status": "scored",
- "ate_4dof": 0.0166,
- "ratio": 0.131,
- "rpe_rot_deg": 0.152,
- "nees_yaw": 0.0015368412884291828,
- "nees_rp": 1.735261870327996,
- "nees_pos": 0.1264184740213932,
- "yaw_sigma_first": 0.10003124796018417,
- "yaw_sigma_min": 0.09470962217689612,
- "yaw_sigma_ratio": 0.9468,
- "yaw_err_max": 0.013668431282133804,
- "nis": 1.00224,
+ "ate_4dof": 0.0238,
+ "ratio": 0.187,
+ "rpe_rot_deg": 1.023,
+ "nees_yaw": 0.05891404594238627,
+ "nees_rp": 12.571429176765228,
+ "nees_pos": 0.11159652053449329,
+ "yaw_sigma_first": 0.1000312479601842,
+ "yaw_sigma_min": 0.06657550530381434,
+ "yaw_sigma_ratio": 0.6655,
+ "yaw_err_max": 0.06767786579456934,
+ "nis": 1.00328,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_01_easy/0.5/1/default": {
"status": "scored",
- "ate_4dof": 0.0196,
- "ratio": 0.155,
- "rpe_rot_deg": 0.248,
- "nees_yaw": 0.004509293133050892,
- "nees_rp": 1.790803337763277,
- "nees_pos": 0.06390070233201936,
- "yaw_sigma_first": 0.10003124796018417,
- "yaw_sigma_min": 0.09533656724701751,
- "yaw_sigma_ratio": 0.9531,
- "yaw_err_max": 0.017469149496301167,
- "nis": 0.0121362,
+ "ate_4dof": 0.045,
+ "ratio": 0.355,
+ "rpe_rot_deg": 1.68,
+ "nees_yaw": 0.633253797729645,
+ "nees_rp": 12.504564970079752,
+ "nees_pos": 0.5297116356208329,
+ "yaw_sigma_first": 0.1000312479601842,
+ "yaw_sigma_min": 0.0679232390816889,
+ "yaw_sigma_ratio": 0.679,
+ "yaw_err_max": 0.07061460218999875,
+ "nis": 0.012556,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_01_easy/0.5/2/matched": {
"status": "scored",
- "ate_4dof": 0.0182,
- "ratio": 0.143,
- "rpe_rot_deg": 0.186,
- "nees_yaw": 0.01264240881194688,
- "nees_rp": 1.1909511604362242,
- "nees_pos": 0.11630041901451617,
- "yaw_sigma_first": 0.10003124796017195,
- "yaw_sigma_min": 0.09490869113758399,
- "yaw_sigma_ratio": 0.9488,
- "yaw_err_max": 0.016866056081276427,
- "nis": 1.00035,
+ "ate_4dof": 0.0172,
+ "ratio": 0.135,
+ "rpe_rot_deg": 0.52,
+ "nees_yaw": 0.007774430286476572,
+ "nees_rp": 2.0366788315255824,
+ "nees_pos": 0.12584606374798016,
+ "yaw_sigma_first": 0.10003124796017199,
+ "yaw_sigma_min": 0.06043741910060418,
+ "yaw_sigma_ratio": 0.6042,
+ "yaw_err_max": 0.018626628293417675,
+ "nis": 1.00088,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_01_easy/0.5/2/default": {
"status": "scored",
- "ate_4dof": 0.034,
- "ratio": 0.268,
- "rpe_rot_deg": 0.29,
- "nees_yaw": 0.02361579542360767,
- "nees_rp": 0.9148762136335052,
- "nees_pos": 0.2832695043273335,
- "yaw_sigma_first": 0.10003124796017195,
- "yaw_sigma_min": 0.09539149930094437,
- "yaw_sigma_ratio": 0.9536,
- "yaw_err_max": 0.01894212447924269,
- "nis": 0.0121234,
+ "ate_4dof": 0.0333,
+ "ratio": 0.263,
+ "rpe_rot_deg": 1.357,
+ "nees_yaw": 0.005363213207041997,
+ "nees_rp": 3.8120344234371952,
+ "nees_pos": 0.2370773350956151,
+ "yaw_sigma_first": 0.10003124796017199,
+ "yaw_sigma_min": 0.06263436462549604,
+ "yaw_sigma_ratio": 0.6261,
+ "yaw_err_max": 0.018626628293417675,
+ "nis": 0.0124214,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_01_easy/0.5/3/matched": {
"status": "scored",
- "ate_4dof": 0.0211,
- "ratio": 0.166,
- "rpe_rot_deg": 0.319,
- "nees_yaw": 0.0020088755911836197,
- "nees_rp": 1.8502902341440584,
- "nees_pos": 0.0929222603583235,
- "yaw_sigma_first": 0.10003124796016664,
- "yaw_sigma_min": 0.09572100114675319,
- "yaw_sigma_ratio": 0.9569,
- "yaw_err_max": 0.007484074233870764,
- "nis": 1.00509,
+ "ate_4dof": 0.0222,
+ "ratio": 0.175,
+ "rpe_rot_deg": 0.727,
+ "nees_yaw": 0.014402641533277583,
+ "nees_rp": 3.750032215089416,
+ "nees_pos": 0.11914750175400096,
+ "yaw_sigma_first": 0.10003124796016667,
+ "yaw_sigma_min": 0.06431178423450604,
+ "yaw_sigma_ratio": 0.6429,
+ "yaw_err_max": 0.016074700668797886,
+ "nis": 1.00599,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_01_easy/0.5/3/default": {
"status": "scored",
- "ate_4dof": 0.0353,
- "ratio": 0.278,
- "rpe_rot_deg": 0.376,
- "nees_yaw": 0.02122980248335341,
- "nees_rp": 1.8938079130015213,
- "nees_pos": 0.1840659085322209,
- "yaw_sigma_first": 0.10003124796016664,
- "yaw_sigma_min": 0.09595200026618075,
- "yaw_sigma_ratio": 0.9592,
- "yaw_err_max": 0.017952840455637754,
- "nis": 0.0122065,
+ "ate_4dof": 0.0417,
+ "ratio": 0.329,
+ "rpe_rot_deg": 1.515,
+ "nees_yaw": 0.0485392490999108,
+ "nees_rp": 5.550736492782289,
+ "nees_pos": 0.211700798449967,
+ "yaw_sigma_first": 0.10003124796016667,
+ "yaw_sigma_min": 0.06636573817163875,
+ "yaw_sigma_ratio": 0.6635,
+ "yaw_err_max": 0.03537427599522237,
+ "nis": 0.0125106,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_02_medium/1/1/matched": {
"status": "scored",
- "ate_4dof": 0.0154,
- "ratio": 0.211,
- "rpe_rot_deg": 0.193,
- "nees_yaw": 0.0011900117979073408,
- "nees_rp": 2.369599299505569,
- "nees_pos": 0.13118342243116754,
- "yaw_sigma_first": 0.10004499291706986,
- "yaw_sigma_min": 0.10000164353530712,
+ "ate_4dof": 0.0172,
+ "ratio": 0.236,
+ "rpe_rot_deg": 0.218,
+ "nees_yaw": 0.0007439998671706955,
+ "nees_rp": 3.2570281292335608,
+ "nees_pos": 0.11372095203916481,
+ "yaw_sigma_first": 0.10004499291706997,
+ "yaw_sigma_min": 0.10000272088593647,
"yaw_sigma_ratio": 0.9996,
- "yaw_err_max": 0.007809525406174077,
- "nis": 1.00095,
+ "yaw_err_max": 0.006950889869248169,
+ "nis": 1.00116,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_02_medium/1/1/default": {
"status": "scored",
- "ate_4dof": 0.0265,
- "ratio": 0.364,
- "rpe_rot_deg": 0.368,
- "nees_yaw": 0.0038062796069786046,
- "nees_rp": 1.634056096138959,
- "nees_pos": 0.23080211973799186,
- "yaw_sigma_first": 0.10004499291706986,
- "yaw_sigma_min": 0.10004499291706986,
- "yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.013258749150166331,
- "nis": 0.047707,
+ "ate_4dof": 0.0437,
+ "ratio": 0.6,
+ "rpe_rot_deg": 0.437,
+ "nees_yaw": 0.016053608589237617,
+ "nees_rp": 5.512047688872166,
+ "nees_pos": 0.30323858525644665,
+ "yaw_sigma_first": 0.10004499291706997,
+ "yaw_sigma_min": 0.10002683248887624,
+ "yaw_sigma_ratio": 0.9998,
+ "yaw_err_max": 0.022011749875323675,
+ "nis": 0.0479673,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_02_medium/1/2/matched": {
"status": "scored",
- "ate_4dof": 0.027,
- "ratio": 0.37,
- "rpe_rot_deg": 0.166,
- "nees_yaw": 0.0006576898425582601,
- "nees_rp": 1.4007376120917754,
- "nees_pos": 0.1117875727022448,
- "yaw_sigma_first": 0.10004499291701974,
- "yaw_sigma_min": 0.10000108351297401,
+ "ate_4dof": 0.0198,
+ "ratio": 0.272,
+ "rpe_rot_deg": 0.258,
+ "nees_yaw": 0.000773112689538334,
+ "nees_rp": 2.6646080769384892,
+ "nees_pos": 0.1370244930871155,
+ "yaw_sigma_first": 0.10004499291701976,
+ "yaw_sigma_min": 0.10000099054111371,
"yaw_sigma_ratio": 0.9996,
- "yaw_err_max": 0.005024212192965161,
- "nis": 1.00154,
+ "yaw_err_max": 0.006159273451357258,
+ "nis": 1.00133,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_02_medium/1/2/default": {
"status": "scored",
- "ate_4dof": 0.0232,
- "ratio": 0.319,
- "rpe_rot_deg": 0.304,
- "nees_yaw": 0.0028125706231609286,
- "nees_rp": 1.057078521543599,
- "nees_pos": 0.0730312916060576,
- "yaw_sigma_first": 0.10004499291701974,
- "yaw_sigma_min": 0.10001777244304744,
- "yaw_sigma_ratio": 0.9997,
- "yaw_err_max": 0.012563041336535515,
- "nis": 0.047744,
+ "ate_4dof": 0.0422,
+ "ratio": 0.579,
+ "rpe_rot_deg": 0.438,
+ "nees_yaw": 0.005241213113995007,
+ "nees_rp": 5.285087913345159,
+ "nees_pos": 0.516599155031004,
+ "yaw_sigma_first": 0.10004499291701976,
+ "yaw_sigma_min": 0.10002227387753124,
+ "yaw_sigma_ratio": 0.9998,
+ "yaw_err_max": 0.014506958543803175,
+ "nis": 0.0480295,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_02_medium/1/3/matched": {
"status": "scored",
- "ate_4dof": 0.0193,
- "ratio": 0.265,
- "rpe_rot_deg": 0.186,
- "nees_yaw": 0.005552296207551066,
- "nees_rp": 1.9815744455605757,
- "nees_pos": 0.2217005054581218,
- "yaw_sigma_first": 0.10004499291696914,
- "yaw_sigma_min": 0.10000637540278802,
+ "ate_4dof": 0.0216,
+ "ratio": 0.297,
+ "rpe_rot_deg": 0.198,
+ "nees_yaw": 0.005873523786925887,
+ "nees_rp": 1.9063540249033841,
+ "nees_pos": 0.08802916021876214,
+ "yaw_sigma_first": 0.10004499291696915,
+ "yaw_sigma_min": 0.10000065575179697,
"yaw_sigma_ratio": 0.9996,
- "yaw_err_max": 0.012633030812898856,
- "nis": 1.00205,
+ "yaw_err_max": 0.012947886022566133,
+ "nis": 1.00373,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_02_medium/1/3/default": {
"status": "scored",
- "ate_4dof": 0.0245,
- "ratio": 0.336,
- "rpe_rot_deg": 0.186,
- "nees_yaw": 0.0011078909323594447,
- "nees_rp": 1.6442537463288531,
- "nees_pos": 0.16154552379447756,
- "yaw_sigma_first": 0.10004499291696914,
- "yaw_sigma_min": 0.10004499291696914,
+ "ate_4dof": 0.0389,
+ "ratio": 0.534,
+ "rpe_rot_deg": 0.265,
+ "nees_yaw": 0.018528037188126015,
+ "nees_rp": 3.289452000202981,
+ "nees_pos": 0.2850295938874654,
+ "yaw_sigma_first": 0.10004499291696915,
+ "yaw_sigma_min": 0.10004499291696915,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.009798669320165982,
- "nis": 0.0478283,
+ "yaw_err_max": 0.0216127615104148,
+ "nis": 0.0480511,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_02_medium/0.5/1/matched": {
"status": "scored",
- "ate_4dof": 0.0091,
- "ratio": 0.125,
- "rpe_rot_deg": 0.129,
- "nees_yaw": 0.004328652384639984,
- "nees_rp": 2.014101951556435,
- "nees_pos": 0.02991438493185277,
- "yaw_sigma_first": 0.10004499291706986,
- "yaw_sigma_min": 0.09453503881648538,
- "yaw_sigma_ratio": 0.9449,
- "yaw_err_max": 0.013364302613395199,
- "nis": 1.00094,
+ "ate_4dof": 0.0194,
+ "ratio": 0.266,
+ "rpe_rot_deg": 0.483,
+ "nees_yaw": 0.19529938072464031,
+ "nees_rp": 11.491841421077364,
+ "nees_pos": 0.03347035594338682,
+ "yaw_sigma_first": 0.10004499291706997,
+ "yaw_sigma_min": 0.07050241364132209,
+ "yaw_sigma_ratio": 0.7047,
+ "yaw_err_max": 0.07478249619611232,
+ "nis": 1.00154,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_02_medium/0.5/1/default": {
"status": "scored",
- "ate_4dof": 0.0213,
- "ratio": 0.293,
- "rpe_rot_deg": 0.217,
- "nees_yaw": 0.010466173702216834,
- "nees_rp": 1.5018391057911042,
- "nees_pos": 0.09339754674087275,
- "yaw_sigma_first": 0.10004499291706986,
- "yaw_sigma_min": 0.09647383834686622,
- "yaw_sigma_ratio": 0.9643,
- "yaw_err_max": 0.017551361985212347,
- "nis": 0.0120865,
+ "ate_4dof": 0.0456,
+ "ratio": 0.626,
+ "rpe_rot_deg": 0.775,
+ "nees_yaw": 1.2339077405581709,
+ "nees_rp": 12.773594823862807,
+ "nees_pos": 0.6273511786018847,
+ "yaw_sigma_first": 0.10004499291706997,
+ "yaw_sigma_min": 0.08030197505243507,
+ "yaw_sigma_ratio": 0.8027,
+ "yaw_err_max": 0.12255201136094035,
+ "nis": 0.0124115,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_02_medium/0.5/2/matched": {
"status": "scored",
- "ate_4dof": 0.0108,
- "ratio": 0.149,
- "rpe_rot_deg": 0.097,
- "nees_yaw": 0.0009788208119435888,
- "nees_rp": 1.2507256549208903,
- "nees_pos": 0.021836649490184355,
- "yaw_sigma_first": 0.10004499291701974,
- "yaw_sigma_min": 0.09380381372386645,
- "yaw_sigma_ratio": 0.9376,
- "yaw_err_max": 0.018747421930842417,
- "nis": 1.00055,
+ "ate_4dof": 0.0129,
+ "ratio": 0.177,
+ "rpe_rot_deg": 0.291,
+ "nees_yaw": 0.059134479552387556,
+ "nees_rp": 2.9668074851657265,
+ "nees_pos": 0.06962506756169157,
+ "yaw_sigma_first": 0.10004499291701976,
+ "yaw_sigma_min": 0.06300420393327806,
+ "yaw_sigma_ratio": 0.6298,
+ "yaw_err_max": 0.11001836534833048,
+ "nis": 1.00059,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_02_medium/0.5/2/default": {
"status": "scored",
- "ate_4dof": 0.0254,
- "ratio": 0.349,
- "rpe_rot_deg": 0.261,
- "nees_yaw": 0.02388755505246724,
- "nees_rp": 1.0191968265917923,
- "nees_pos": 0.1499344512221635,
- "yaw_sigma_first": 0.10004499291701974,
- "yaw_sigma_min": 0.09564741093934981,
- "yaw_sigma_ratio": 0.956,
- "yaw_err_max": 0.019662443915867033,
- "nis": 0.0121134,
+ "ate_4dof": 0.0309,
+ "ratio": 0.425,
+ "rpe_rot_deg": 0.403,
+ "nees_yaw": 0.20462149974282776,
+ "nees_rp": 3.053211022736709,
+ "nees_pos": 0.38822300539757937,
+ "yaw_sigma_first": 0.10004499291701976,
+ "yaw_sigma_min": 0.06903511199793796,
+ "yaw_sigma_ratio": 0.69,
+ "yaw_err_max": 0.039258033812730395,
+ "nis": 0.0121874,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_02_medium/0.5/3/matched": {
"status": "scored",
- "ate_4dof": 0.0122,
- "ratio": 0.168,
- "rpe_rot_deg": 0.095,
- "nees_yaw": 0.0026704548845929246,
- "nees_rp": 1.98683181849159,
- "nees_pos": 0.061858137126972995,
- "yaw_sigma_first": 0.10004499291696914,
- "yaw_sigma_min": 0.09453483601206338,
- "yaw_sigma_ratio": 0.9449,
- "yaw_err_max": 0.023760567209205656,
- "nis": 1.00256,
+ "ate_4dof": 0.0118,
+ "ratio": 0.162,
+ "rpe_rot_deg": 0.17,
+ "nees_yaw": 0.34491517551105233,
+ "nees_rp": 2.2808939867399323,
+ "nees_pos": 0.09040324400544576,
+ "yaw_sigma_first": 0.10004499291696915,
+ "yaw_sigma_min": 0.06742288845144923,
+ "yaw_sigma_ratio": 0.6739,
+ "yaw_err_max": 0.050592008399437366,
+ "nis": 1.0027,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_02_medium/0.5/3/default": {
"status": "scored",
- "ate_4dof": 0.0226,
- "ratio": 0.311,
- "rpe_rot_deg": 0.133,
- "nees_yaw": 0.0018600967959766473,
- "nees_rp": 1.7829325885786522,
- "nees_pos": 0.11492197117233587,
- "yaw_sigma_first": 0.10004499291696914,
- "yaw_sigma_min": 0.09659428480771184,
- "yaw_sigma_ratio": 0.9655,
- "yaw_err_max": 0.018739092601538672,
- "nis": 0.0121584,
+ "ate_4dof": 0.0277,
+ "ratio": 0.381,
+ "rpe_rot_deg": 0.349,
+ "nees_yaw": 0.04592240265912469,
+ "nees_rp": 2.2627238906952933,
+ "nees_pos": 0.17573454141758368,
+ "yaw_sigma_first": 0.10004499291696915,
+ "yaw_sigma_min": 0.07591751469557735,
+ "yaw_sigma_ratio": 0.7588,
+ "yaw_err_max": 0.029445752947911387,
+ "nis": 0.0122265,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_03_difficult/1/1/matched": {
"status": "scored",
- "ate_4dof": 0.0437,
- "ratio": 0.242,
- "rpe_rot_deg": 0.161,
- "nees_yaw": 0.0019229712891864542,
- "nees_rp": 1.968740593077678,
- "nees_pos": 0.2581738186923106,
- "yaw_sigma_first": 0.10002000052174864,
- "yaw_sigma_min": 0.1000034064113303,
+ "ate_4dof": 0.0286,
+ "ratio": 0.158,
+ "rpe_rot_deg": 0.181,
+ "nees_yaw": 0.0008102769642167386,
+ "nees_rp": 2.190782352524244,
+ "nees_pos": 0.28206062421334643,
+ "yaw_sigma_first": 0.10002000052174868,
+ "yaw_sigma_min": 0.10000407728147181,
"yaw_sigma_ratio": 0.9998,
- "yaw_err_max": 0.012556077580571772,
- "nis": 1.01231,
+ "yaw_err_max": 0.008581490587239374,
+ "nis": 1.00475,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_03_difficult/1/1/default": {
"status": "scored",
- "ate_4dof": 0.0466,
- "ratio": 0.258,
- "rpe_rot_deg": 0.28,
- "nees_yaw": 0.003987889642693989,
- "nees_rp": 1.4363255598397564,
- "nees_pos": 0.06322343934819365,
- "yaw_sigma_first": 0.10002000052174864,
- "yaw_sigma_min": 0.10002000052174864,
+ "ate_4dof": 0.0356,
+ "ratio": 0.197,
+ "rpe_rot_deg": 0.336,
+ "nees_yaw": 0.00267805191811873,
+ "nees_rp": 2.2296315593253473,
+ "nees_pos": 0.33748132355936417,
+ "yaw_sigma_first": 0.10002000052174868,
+ "yaw_sigma_min": 0.10002000052174868,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.011155476855969524,
- "nis": 0.047785,
+ "yaw_err_max": 0.010496537147172393,
+ "nis": 0.0479246,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_03_difficult/1/2/matched": {
"status": "scored",
- "ate_4dof": 0.2409,
- "ratio": 1.334,
- "rpe_rot_deg": 0.171,
- "nees_yaw": 0.003916265383721754,
- "nees_rp": 1.918064956693952,
- "nees_pos": 0.7136800938314098,
- "yaw_sigma_first": 0.10002000052174878,
- "yaw_sigma_min": 0.10000112034491367,
+ "ate_4dof": 0.0236,
+ "ratio": 0.131,
+ "rpe_rot_deg": 0.223,
+ "nees_yaw": 0.002703060866389148,
+ "nees_rp": 1.8177970502022294,
+ "nees_pos": 0.19519634629199553,
+ "yaw_sigma_first": 0.1000200005217488,
+ "yaw_sigma_min": 0.10000080117617724,
"yaw_sigma_ratio": 0.9998,
- "yaw_err_max": 0.010267447265033277,
- "nis": 1.0064,
+ "yaw_err_max": 0.009750673324149414,
+ "nis": 1.00103,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_03_difficult/1/2/default": {
"status": "scored",
- "ate_4dof": 0.0301,
- "ratio": 0.167,
- "rpe_rot_deg": 0.297,
- "nees_yaw": 0.013778319471874587,
- "nees_rp": 0.8979005714711095,
- "nees_pos": 0.11005641409375179,
- "yaw_sigma_first": 0.10002000052174878,
- "yaw_sigma_min": 0.100015759972545,
+ "ate_4dof": 0.0333,
+ "ratio": 0.185,
+ "rpe_rot_deg": 0.365,
+ "nees_yaw": 0.011575846313698441,
+ "nees_rp": 1.7692229063158371,
+ "nees_pos": 0.26920328658580334,
+ "yaw_sigma_first": 0.1000200005217488,
+ "yaw_sigma_min": 0.1000200005217488,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.024846127966439643,
- "nis": 0.0477753,
+ "yaw_err_max": 0.023133444954206862,
+ "nis": 0.0479763,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_03_difficult/1/3/matched": {
"status": "scored",
- "ate_4dof": 0.0192,
- "ratio": 0.106,
- "rpe_rot_deg": 0.2,
- "nees_yaw": 0.005621688938698282,
- "nees_rp": 1.9561994489196437,
- "nees_pos": 0.23321222126407107,
- "yaw_sigma_first": 0.10002000052174181,
- "yaw_sigma_min": 0.10000198984743891,
+ "ate_4dof": 0.0188,
+ "ratio": 0.104,
+ "rpe_rot_deg": 0.28,
+ "nees_yaw": 0.003370843452255791,
+ "nees_rp": 2.3076935167232415,
+ "nees_pos": 0.1895918978957913,
+ "yaw_sigma_first": 0.10002000052174183,
+ "yaw_sigma_min": 0.10000338720339706,
"yaw_sigma_ratio": 0.9998,
- "yaw_err_max": 0.011800316722957192,
- "nis": 1.00293,
+ "yaw_err_max": 0.01019119774479304,
+ "nis": 1.00344,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_03_difficult/1/3/default": {
"status": "scored",
- "ate_4dof": 0.0497,
- "ratio": 0.275,
- "rpe_rot_deg": 0.174,
- "nees_yaw": 0.0014300353950031592,
- "nees_rp": 1.413620544845129,
- "nees_pos": 0.09587206423831461,
- "yaw_sigma_first": 0.10002000052174181,
- "yaw_sigma_min": 0.10002000052174181,
+ "ate_4dof": 0.0429,
+ "ratio": 0.238,
+ "rpe_rot_deg": 0.221,
+ "nees_yaw": 0.0011323094930810042,
+ "nees_rp": 2.239013273842093,
+ "nees_pos": 1.9864768304462295,
+ "yaw_sigma_first": 0.10002000052174183,
+ "yaw_sigma_min": 0.10002000052174183,
"yaw_sigma_ratio": 1.0,
- "yaw_err_max": 0.008093032772085815,
- "nis": 0.0479385,
+ "yaw_err_max": 0.010220887222758285,
+ "nis": 0.048,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_03_difficult/0.5/1/matched": {
"status": "scored",
- "ate_4dof": 0.0127,
- "ratio": 0.07,
- "rpe_rot_deg": 0.101,
- "nees_yaw": 0.025504858657117473,
- "nees_rp": 1.9631040441743737,
- "nees_pos": 0.030248311657825728,
- "yaw_sigma_first": 0.10002000052174864,
- "yaw_sigma_min": 0.0956798440169775,
- "yaw_sigma_ratio": 0.9566,
- "yaw_err_max": 0.021098237232488007,
- "nis": 1.00189,
+ "ate_4dof": 0.013,
+ "ratio": 0.072,
+ "rpe_rot_deg": 0.253,
+ "nees_yaw": 0.005067261877560155,
+ "nees_rp": 5.065170322248902,
+ "nees_pos": 0.045993650060657185,
+ "yaw_sigma_first": 0.10002000052174868,
+ "yaw_sigma_min": 0.08225725090520208,
+ "yaw_sigma_ratio": 0.8224,
+ "yaw_err_max": 0.05215926552267274,
+ "nis": 1.00198,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_03_difficult/0.5/1/default": {
"status": "scored",
- "ate_4dof": 0.017,
- "ratio": 0.094,
- "rpe_rot_deg": 0.128,
- "nees_yaw": 0.0025030675197585087,
- "nees_rp": 1.4260120135566559,
- "nees_pos": 0.045470659947827126,
- "yaw_sigma_first": 0.10002000052174864,
- "yaw_sigma_min": 0.09753170215409397,
- "yaw_sigma_ratio": 0.9751,
- "yaw_err_max": 0.014233337936349563,
- "nis": 0.0120946,
+ "ate_4dof": 0.0222,
+ "ratio": 0.123,
+ "rpe_rot_deg": 0.35,
+ "nees_yaw": 0.3709927199337421,
+ "nees_rp": 3.091332903055369,
+ "nees_pos": 0.14716169254944705,
+ "yaw_sigma_first": 0.10002000052174868,
+ "yaw_sigma_min": 0.08509136226536651,
+ "yaw_sigma_ratio": 0.8507,
+ "yaw_err_max": 0.05926399715412647,
+ "nis": 0.0122104,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_03_difficult/0.5/2/matched": {
"status": "scored",
- "ate_4dof": 0.0131,
+ "ate_4dof": 0.0132,
"ratio": 0.073,
- "rpe_rot_deg": 0.1,
- "nees_yaw": 0.00018220600229076043,
- "nees_rp": 1.451455700037378,
- "nees_pos": 0.04200569777184287,
- "yaw_sigma_first": 0.10002000052174878,
- "yaw_sigma_min": 0.0970551938460797,
- "yaw_sigma_ratio": 0.9704,
- "yaw_err_max": 0.006544162726205793,
- "nis": 1.00035,
+ "rpe_rot_deg": 0.131,
+ "nees_yaw": 0.07819659222618143,
+ "nees_rp": 1.536796511461853,
+ "nees_pos": 0.06106038472694937,
+ "yaw_sigma_first": 0.1000200005217488,
+ "yaw_sigma_min": 0.07997965654257433,
+ "yaw_sigma_ratio": 0.7996,
+ "yaw_err_max": 0.026617940097116946,
+ "nis": 1.00025,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_03_difficult/0.5/2/default": {
"status": "scored",
- "ate_4dof": 0.0198,
- "ratio": 0.11,
- "rpe_rot_deg": 0.217,
- "nees_yaw": 0.0060705419871568056,
- "nees_rp": 0.9543255977258005,
- "nees_pos": 0.067488216729898,
- "yaw_sigma_first": 0.10002000052174878,
- "yaw_sigma_min": 0.09802854542499063,
- "yaw_sigma_ratio": 0.9801,
- "yaw_err_max": 0.017994055319151797,
- "nis": 0.0121268,
+ "ate_4dof": 0.019,
+ "ratio": 0.105,
+ "rpe_rot_deg": 0.262,
+ "nees_yaw": 0.19553943111926148,
+ "nees_rp": 1.6369603052097697,
+ "nees_pos": 0.07814606215340517,
+ "yaw_sigma_first": 0.1000200005217488,
+ "yaw_sigma_min": 0.08268587179417347,
+ "yaw_sigma_ratio": 0.8267,
+ "yaw_err_max": 0.049285718394854676,
+ "nis": 0.0121811,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_03_difficult/0.5/3/matched": {
"status": "scored",
- "ate_4dof": 0.0082,
- "ratio": 0.045,
- "rpe_rot_deg": 0.108,
- "nees_yaw": 0.0012889986468632506,
- "nees_rp": 1.8749950484862121,
- "nees_pos": 0.009967861966303573,
- "yaw_sigma_first": 0.10002000052174181,
- "yaw_sigma_min": 0.09711738535820791,
- "yaw_sigma_ratio": 0.971,
- "yaw_err_max": 0.009958728786649867,
- "nis": 1.00347,
+ "ate_4dof": 0.0083,
+ "ratio": 0.046,
+ "rpe_rot_deg": 0.138,
+ "nees_yaw": 0.1541408123694739,
+ "nees_rp": 1.9671917410257702,
+ "nees_pos": 0.012414373801198595,
+ "yaw_sigma_first": 0.10002000052174183,
+ "yaw_sigma_min": 0.08187186791357694,
+ "yaw_sigma_ratio": 0.8186,
+ "yaw_err_max": 0.03494077402583822,
+ "nis": 1.0035,
"ledger_info_yaw": null,
"ledger_share_s6": null
},
"V2_03_difficult/0.5/3/default": {
"status": "scored",
- "ate_4dof": 0.0217,
- "ratio": 0.12,
- "rpe_rot_deg": 0.103,
- "nees_yaw": 0.0011383870949396293,
- "nees_rp": 1.3561636226954683,
- "nees_pos": 0.07123189579999492,
- "yaw_sigma_first": 0.10002000052174181,
- "yaw_sigma_min": 0.09818822976184617,
- "yaw_sigma_ratio": 0.9817,
- "yaw_err_max": 0.009958728786649867,
- "nis": 0.0121783,
+ "ate_4dof": 0.0235,
+ "ratio": 0.13,
+ "rpe_rot_deg": 0.164,
+ "nees_yaw": 0.1389346230336542,
+ "nees_rp": 1.5577804634408843,
+ "nees_pos": 0.07134335900052698,
+ "yaw_sigma_first": 0.10002000052174183,
+ "yaw_sigma_min": 0.08486776916565124,
+ "yaw_sigma_ratio": 0.8485,
+ "yaw_err_max": 0.036053321445847795,
+ "nis": 0.0122231,
"ledger_info_yaw": null,
"ledger_share_s6": null
}
diff --git a/sdk/include/branes/sdk/msckf/stages/s1_initialization.hpp b/sdk/include/branes/sdk/msckf/stages/s1_initialization.hpp
index 066fff8..6eefa57 100644
--- a/sdk/include/branes/sdk/msckf/stages/s1_initialization.hpp
+++ b/sdk/include/branes/sdk/msckf/stages/s1_initialization.hpp
@@ -3,8 +3,9 @@
// branes/sdk/msckf/stages/s1_initialization.hpp — stage S1_initialization as an
// explicit transformation (issue #452, docs/pipeline/canonical-pipeline.md §S1).
//
-// seed_from_imu(state, init result, t, accel window) → (state', diagnostics)
+// seed_from_imu(state, init result, t, accel window, prior) → (state', diagnostics)
// seed_from_alignment(state, VI alignment, SfM window, t, ...) → (state', diagnostics)
+// tilt_bias_prior(accel window, method, σ_ba, ...) → what a static window determines (#609)
//
// Seeds a fresh filter state from a resolved initialization: the static /
// gravity-only / identity bootstrap of ImuInitializer, or the dynamic
@@ -22,6 +23,7 @@
#include
#include
+#include
#include
namespace branes::sdk {
@@ -100,6 +102,152 @@ struct InitDiagnostics {
namespace msckf::stages::s1_initialization {
+/// What an IMU seed window determines about tilt and accelerometer bias (#609).
+///
+/// At rest the accelerometer reads f = Rᵀ·g·e_z + b_a + n. A seed that takes
+/// its "up" û from the window's mean f̄ with b_a = 0 makes the tilt error
+/// δθ (R = R̂·Exp(δθ)) and the bias b_a one quantity seen once:
+///
+/// δθ⊥ = [û]ₓ (b_a + n̄) / g, the yaw component δθ·û untouched.
+///
+/// So P₀ is not isotropic. With b_a ~ N(0, σ_ba² I) and the mean force's own
+/// error n̄ ~ N(0, σ_f² I):
+///
+/// P_θθ = (σ_ba² + σ_f²)/g² · (I − ûûᵀ) + σ_ψ² ûûᵀ
+/// P_θb = σ_ba²/g · [û]ₓ (P_bθ = P_θbᵀ)
+/// P_bb = σ_ba² (I − ûûᵀ) + σ_∥² ûûᵀ
+///
+/// The tilt and the horizontal bias are nearly fully correlated: the filter
+/// starts knowing their combination, not either alone, and separates them as
+/// the platform rotates. Along û the window sees |f̄| − g = b_a·û + n̄·û: a
+/// static seed takes the bias along up from it, with σ_∥ the error of that
+/// difference; otherwise σ_∥ = σ_ba.
+template
+struct TiltBiasPrior {
+ T accel_bias_sigma = T{0}; ///< σ_ba per axis; 0 leaves the fresh state's prior as it is
+ T mean_force_sigma = T{0}; ///< σ_f: per-axis error of the window's mean specific force
+ T along_up_sigma = T{0}; ///< σ_∥ when the bias along up is seeded from |f̄| − g; 0 = not seeded
+};
+
+/// Half the spread of surface gravity (9.780 at the equator to 9.832 m/s² at the
+/// poles): the error of a configured |g| that was not measured on site. It
+/// bounds how well |f̄| − g measures the bias along up.
+template
+inline constexpr T kGravityMagnitudeSigma = T{26} / T{1000};
+
+/// The prior a seed window implies, for `method`:
+/// - Static: the window is at rest, so σ_f is its mean's noise, the larger of
+/// the sample scatter over √N and the white-noise model σ_a/√(window); the
+/// bias along up is seeded, σ_∥² = σ_f² + kGravityMagnitudeSigma².
+/// - GravityAlign, Dynamic and Identity: none; the fresh state's isotropic
+/// prior stays. A gravity-aligned window moves: its mean force includes a
+/// mean acceleration (|f̄| − g is 2.6 m/s² on the EuRoC gate's MH_03 seeds),
+/// so the honest tilt σ is about 15°, and the seed's zero velocity (σ 0.1
+/// m/s) is then what is wrong. Given that tilt σ the first updates explain
+/// the velocity by tilt (2.4° → 12–19°) and three of the gate's five MH_03
+/// draws diverge, against one with the isotropic 5.7° (#609). A dynamic
+/// seed's tilt comes from the visual-inertial alignment, not the
+/// accelerometer mean.
+/// `accel_bias_sigma` ≤ 0 disables the prior (the isotropic P₀ before #609).
+template
+[[nodiscard]] TiltBiasPrior tilt_bias_prior(std::span> accel_window,
+ InitMethod method,
+ T accel_bias_sigma,
+ T accel_noise_density,
+ T window_seconds) {
+ namespace ld = math::lie::detail;
+ using DVec3 = ld::Vec;
+ TiltBiasPrior p;
+ const std::size_t n = accel_window.size();
+ if (!(accel_bias_sigma > T{0}) || n < 2 || method != InitMethod::Static)
+ return p;
+ DVec3 am{};
+ for (const DVec3& a : accel_window)
+ am = am + a;
+ am = am * (T{1} / static_cast(n));
+ T ss{0};
+ for (const DVec3& a : accel_window) {
+ const DVec3 d = a - am;
+ ss = ss + ld::dot(d, d);
+ }
+ const T scatter2 = ss / (T{3} * static_cast(n - 1)); // per-axis sample variance
+ const T model2 = window_seconds > T{0} ? accel_noise_density * accel_noise_density / window_seconds : T{0};
+ const T mean2 = scatter2 / static_cast(n) > model2 ? scatter2 / static_cast(n) : model2;
+ p.accel_bias_sigma = accel_bias_sigma;
+ p.mean_force_sigma = ld::sqrt_(mean2);
+ p.along_up_sigma = ld::sqrt_(mean2 + kGravityMagnitudeSigma * kGravityMagnitudeSigma);
+ return p;
+}
+
+/// Rewrite the attitude and accelerometer-bias blocks of a fresh state's prior
+/// with `prior`, about the seed's body "up" û = R̂ᵀe_z (see TiltBiasPrior). The
+/// yaw variance and every other block are kept. `s` must be fresh: no clones,
+/// and no IMU–calibration cross-covariance. Returns false, leaving `s` as it
+/// was, when the prior is disabled or not positive-definite.
+template
+bool apply_tilt_bias_prior(State& s, T gravity_magnitude, const TiltBiasPrior& prior) {
+ namespace ld = math::lie::detail;
+ using St = State;
+ constexpr std::size_t n = St::kImuDim;
+ if (!(prior.accel_bias_sigma > T{0}) || !(gravity_magnitude > T{0}))
+ return false;
+ const auto ez = ld::Vec{{T{0}, T{0}, T{1}}};
+ const auto u = s.R.inverse() * ez;
+ const auto ux = ld::hat(u);
+ const DynMat P = s.covariance();
+ DynMat cur(n, n), M(n, n);
+ for (std::size_t i = 0; i < n; ++i)
+ for (std::size_t j = 0; j < n; ++j)
+ cur(i, j) = M(i, j) = P(i, j);
+ T yaw2{0}; // the yaw variance ûᵀ P_θθ û, kept
+ for (std::size_t i = 0; i < 3; ++i)
+ for (std::size_t j = 0; j < 3; ++j)
+ yaw2 = yaw2 + u[i] * P(St::kTheta + i, St::kTheta + j) * u[j];
+ const T sb2 = prior.accel_bias_sigma * prior.accel_bias_sigma;
+ const T sf2 = prior.mean_force_sigma * prior.mean_force_sigma;
+ const T g = gravity_magnitude;
+ const T tilt2 = (sb2 + sf2) / (g * g);
+ const T par2 = prior.along_up_sigma > T{0} ? prior.along_up_sigma * prior.along_up_sigma : sb2;
+ for (std::size_t i = 0; i < 3; ++i) {
+ for (std::size_t j = 0; j < 3; ++j) {
+ const T uu = u[i] * u[j];
+ const T perp = (i == j ? T{1} : T{0}) - uu;
+ M(St::kTheta + i, St::kTheta + j) = tilt2 * perp + yaw2 * uu;
+ M(St::kBa + i, St::kBa + j) = sb2 * perp + par2 * uu;
+ M(St::kTheta + i, St::kBa + j) = sb2 / g * ux(i, j);
+ M(St::kBa + j, St::kTheta + i) = sb2 / g * ux(i, j);
+ }
+ for (std::size_t k = 0; k < n; ++k) { // nothing else is correlated with the new blocks
+ if (k < St::kTheta || k >= St::kTheta + 3) {
+ if (k < St::kBa || k >= St::kBa + 3) {
+ M(St::kTheta + i, k) = M(k, St::kTheta + i) = T{0};
+ M(St::kBa + i, k) = M(k, St::kBa + i) = T{0};
+ }
+ }
+ }
+ }
+ // P' = G P Gᵀ with G = L_M · L_cur⁻¹ on the IMU block, identity elsewhere.
+ DynMat Lm, Lc;
+ if (!cholesky(M, Lm) || !cholesky(cur, Lc))
+ return false;
+ DynMat Lci(n, n); // L_cur⁻¹ by forward substitution, column by column
+ for (std::size_t c = 0; c < n; ++c) {
+ for (std::size_t i = c; i < n; ++i) {
+ T acc = i == c ? T{1} : T{0};
+ for (std::size_t k = c; k < i; ++k)
+ acc = acc - Lc(i, k) * Lci(k, c);
+ Lci(i, c) = acc / Lc(i, i);
+ }
+ }
+ const DynMat Gi = mul(Lm, Lci);
+ DynMat G = DynMat::identity(s.cov.dim());
+ for (std::size_t i = 0; i < n; ++i)
+ for (std::size_t j = 0; j < n; ++j)
+ G(i, j) = Gi(i, j);
+ s.cov.transform(G);
+ return true;
+}
+
/// Seed `s` (a fresh state) from an IMU-only initialization — static, gravity
/// alignment, or the identity fallback (a default-constructed `res`) — and
/// record the result in `diag`: method, time, the measured up-direction and
@@ -112,7 +260,8 @@ void seed_from_imu(State& s,
double t,
std::span> accel_window,
T gravity_magnitude,
- InitDiagnostics& diag) {
+ InitDiagnostics& diag,
+ const TiltBiasPrior& prior = {}) {
using DVec3 = math::lie::detail::Vec;
s.R = res.R_world_imu; // identity for a default-constructed result
s.bg = res.gyro_bias;
@@ -132,8 +281,14 @@ void seed_from_imu(State& s,
// Explicit narrowing: the diagnostic is double telemetry, T may be a posit.
diag.gravity_residual =
static_cast((g_meas > T{0} && g_cfg > T{0}) ? math::lie::detail::abs_(g_meas - g_cfg) / g_cfg : T{1});
+ // The bias along up, from |f̄| − g, when the prior says the window resolves it.
+ if (prior.along_up_sigma > T{0} && g_meas > T{0}) {
+ const DVec3 up = s.R.inverse() * DVec3{{T{0}, T{0}, T{1}}};
+ s.ba = s.ba + up * (g_meas - g_cfg);
+ }
+ (void)apply_tilt_bias_prior(s, g_cfg, prior);
diag.gyro_bias = res.gyro_bias;
- diag.accel_bias = res.accel_bias;
+ diag.accel_bias = s.ba;
}
/// Seed `s` (a fresh state) from a successful dynamic visual-inertial
diff --git a/sdk/include/branes/sdk/msckf_backend.hpp b/sdk/include/branes/sdk/msckf_backend.hpp
index 3c493b9..078f247 100644
--- a/sdk/include/branes/sdk/msckf_backend.hpp
+++ b/sdk/include/branes/sdk/msckf_backend.hpp
@@ -170,6 +170,7 @@ class MsckfBackendT final : public VioBackend {
last_sample_t_ = 0.0;
init_gyro_.clear();
init_accel_.clear();
+ init_t_.clear();
init_samples_seen_ = 0;
rest_bias_.reset();
init_frames_.clear();
@@ -589,6 +590,7 @@ class MsckfBackendT final : public VioBackend {
void try_initialize(const DVec3& gyro, const DVec3& accel, double t) {
init_gyro_.push_back(gyro);
init_accel_.push_back(accel);
+ init_t_.push_back(t);
++init_samples_seen_;
// Slide a fixed window of the most recent kInitSamples: the static
// check wants the latest stretch (most likely to catch a quiet
@@ -599,6 +601,7 @@ class MsckfBackendT final : public VioBackend {
if (init_gyro_.size() > kInitSamples) {
init_gyro_.erase(init_gyro_.begin());
init_accel_.erase(init_accel_.begin());
+ init_t_.erase(init_t_.begin());
}
if (init_gyro_.size() < kInitSamples)
return;
@@ -659,16 +662,32 @@ class MsckfBackendT final : public VioBackend {
t,
std::span{init_accel_},
static_cast(config_.gravity_magnitude),
- init_diag_);
+ init_diag_,
+ seed_prior(method));
finish_init(t);
}
+ // S1: the tilt–accelerometer-bias prior the IMU seed window implies (#609).
+ [[nodiscard]] msckf::stages::s1_initialization::TiltBiasPrior seed_prior(InitMethod method) const {
+ // The window spans N samples, so N − 1 intervals of the IMU's period.
+ const double span = init_t_.size() > 1
+ ? (init_t_.back() - init_t_.front()) * static_cast(init_t_.size()) /
+ static_cast(init_t_.size() - 1)
+ : 0.0;
+ return msckf::stages::s1_initialization::tilt_bias_prior(std::span{init_accel_},
+ method,
+ static_cast(config_.accel_bias_prior_sigma),
+ static_cast(config_.accel_noise_density),
+ static_cast(span));
+ }
+
void finish_init(double t) {
initialized_ = true;
have_last_time_ = true;
last_imu_time_ = t;
init_gyro_.clear();
init_accel_.clear();
+ init_t_.clear();
init_frames_.clear();
}
@@ -901,6 +920,7 @@ class MsckfBackendT final : public VioBackend {
std::vector init_gyro_;
std::optional> rest_bias_; ///< stillest vibrating_rest_bias while waiting to init (#591)
std::vector init_accel_;
+ std::vector init_t_; ///< the init window's sample times
std::size_t init_samples_seen_ = 0; ///< total IMU samples seen during init (for the timeout)
// Dynamic-init state: a bounded buffer of init-window camera frames and the
diff --git a/sdk/include/branes/sdk/vio_backend.hpp b/sdk/include/branes/sdk/vio_backend.hpp
index f5f1cf0..7d2d43f 100644
--- a/sdk/include/branes/sdk/vio_backend.hpp
+++ b/sdk/include/branes/sdk/vio_backend.hpp
@@ -42,6 +42,12 @@ struct VioConfig {
double gyro_noise_density = 1.6e-4; ///< continuous-time σ_g
double accel_bias_random_walk = 3.0e-3;
double gyro_bias_random_walk = 1.9e-5;
+ /// S1 prior σ of the accelerometer bias at the seed, m/s² per axis (#609).
+ /// A static window measures tilt and horizontal bias only as one sum, so
+ /// the seed's P₀ correlates them (σ_tilt ≈ σ_ba / g) and the filter learns
+ /// the bias as the platform rotates. 0.2 puts EuRoC's largest ground-truth
+ /// bias (V1_01, 0.55 m/s²) inside 3σ. 0 restores the isotropic P₀.
+ double accel_bias_prior_sigma = 0.2;
/// Visual measurement σ in normalized image coordinates — the MSCKF camera
/// update's R. Default matches the updater's historical hardcoded value.
double camera_noise_normalized = 1.0e-2;
diff --git a/tests/sdk/calibration_probe.cpp b/tests/sdk/calibration_probe.cpp
index 45b4a35..614d849 100644
--- a/tests/sdk/calibration_probe.cpp
+++ b/tests/sdk/calibration_probe.cpp
@@ -50,16 +50,22 @@ TEST_CASE("S10 end-to-end: with perfect calibration and exact sensors the world
const auto sw = ev::r_inflation_sweep();
REQUIRE(sw.curve.size() >= 5);
// With PERFECT calibration and EXACT sensors the pose NEES is below the
- // 6-DoF target (0.04 measured) and the pose tracks to centimetres. The
+ // 6-DoF target (0.007 measured; 0.04 before the #609 seed prior) and the
+ // pose tracks to centimetres. The
// assumed R (4.6 px) is pessimistic against zero noise, so this checks the
// world and the plumbing, not consistency under noise: that is the
// matched-noise control below (#551).
REQUIRE(sw.curve.front().r_scale == 1.0);
REQUIRE(sw.curve.front().nees < 6.0);
REQUIRE(sw.curve.front().ate < 0.05);
- // Inflating R only makes the filter less confident…
- for (std::size_t i = 1; i < sw.curve.size(); ++i)
- REQUIRE(sw.curve[i].nees <= sw.curve[i - 1].nees);
+ // Inflating R never makes the filter over-confident. With exact sensors the
+ // error grows with R about as fast as the covariance (NEES 0.007-0.009 from
+ // R x1 to x64 since #609), so the NEES is flat, not falling: what holds at
+ // every scale is that it stays under the dof.
+ for (const auto& pt : sw.curve) {
+ INFO("R x" << pt.r_scale << ": NEES " << pt.nees);
+ REQUIRE(pt.nees < 6.0);
+ }
// …and no R-scale is needed to reach consistency.
REQUIRE(sw.r_scale_for_consistency == 0.0);
// The 1° calibration budget is a real fraction of that deficit (the R×4 part).
@@ -84,7 +90,7 @@ TEST_CASE("S10 matched-noise control: the body-frame MSCKF leaks information int
CHECK(r.ate < 0.5); // it still tracks: the defect is the covariance
// The leak signature: an unobservable direction gained information.
CHECK(r.yaw_sigma_last < r.yaw_sigma_first / 3);
- nees.push_back(r.nees);
+ nees.push_back(r.nees_yaw);
pos.push_back(r.nees_pos);
}
// Position is gauge too, but its σ stays near the prior and is mostly honest:
@@ -93,9 +99,12 @@ TEST_CASE("S10 matched-noise control: the body-frame MSCKF leaks information int
// is on the median, 2× the dof.
std::sort(pos.begin(), pos.end());
CHECK(pos[pos.size() / 2] < 6.0);
- // Median 6-DoF pose NEES; consistent ≈ 6. Seeds 1–3 measure 23.5, 10.6 and
- // 112; over seeds 1–10 it is 10.6–112, median 34 (11.5–288, median 47, under
- // the 5-per-dof gate before #481). The bound asks for at least 1.5× the dof.
+ // Median yaw NEES; consistent ≈ 1. Seeds 1–3 measure 5.6, 2.0 and 112. The
+ // bound asks for at least 1.5× the dof. Until #609 this was the 6-DoF pose
+ // NEES (seeds 1–3: 23.5, 10.6, 112), which also carried the seed's
+ // roll/pitch over-confidence; with the window's tilt-bias prior roll/pitch
+ // read 1.8, 2.5 and 15 against 2, the pose NEES 7.3, 4.6 and 143, and what
+ // is left over-confident is the yaw.
std::sort(nees.begin(), nees.end());
- CHECK(nees[nees.size() / 2] > 9.0);
+ CHECK(nees[nees.size() / 2] > 1.5);
}
diff --git a/tests/sdk/tilt_bias_prior.cpp b/tests/sdk/tilt_bias_prior.cpp
new file mode 100644
index 0000000..2fad8a1
--- /dev/null
+++ b/tests/sdk/tilt_bias_prior.cpp
@@ -0,0 +1,362 @@
+// The S1 seed's tilt-accelerometer-bias prior (issue #609).
+//
+// A static window measures the accelerometer's mean specific force
+// f = R^T g e_z + b_a. Taking "up" from it with b_a = 0 makes the tilt error and
+// the horizontal bias one quantity, seen once: the seed must say so in P0
+// (stages::s1_initialization::TiltBiasPrior), and the filter then learns the
+// bias as the platform rotates. Before #609 P0 was an isotropic sigma = 0.1
+// with no correlation, and a 0.55 m/s^2 bias (EuRoC V1_01) left both filters
+// tilted by 2.4-2.6 deg for the whole run.
+//
+// ASCII-only TEST_CASE names (the Windows MSVC ctest job rejects non-ASCII).
+
+#include
+#include
+#include
+#include
+#include
+#include
+#include
+
+#include
+
+#include
+#include
+#include
+#include
+#include
+#include
+#include
+
+namespace {
+
+namespace bs = branes::sdk;
+namespace ms = branes::sdk::msckf;
+namespace s1 = branes::sdk::msckf::stages::s1_initialization;
+namespace ld = branes::math::lie::detail;
+using T = double;
+using Vec3 = ld::Vec;
+using SO3 = branes::math::lie::SO3;
+using State = ms::State;
+
+constexpr T kG = 9.81;
+constexpr T kRate = 200.0; // Hz
+constexpr std::size_t kN = 50; // the backend's static window
+constexpr T kDensity = 2.0e-3; // VioConfig::accel_noise_density
+constexpr T kSigmaBa = 0.2; // VioConfig::accel_bias_prior_sigma
+
+// Gaussian draws from the engine's raw output (Box-Muller): std::normal_distribution
+// is implementation-defined, and the draws must match across platforms (#551).
+struct Gauss {
+ std::mt19937_64 eng;
+ explicit Gauss(std::uint64_t seed) : eng(seed) {}
+ double operator()() {
+ constexpr double k = 1.0 / 18446744073709551616.0; // 2^-64
+ const double u1 = (static_cast(eng()) + 0.5) * k;
+ const double u2 = (static_cast(eng()) + 0.5) * k;
+ return std::sqrt(-2.0 * std::log(u1)) * std::cos(6.283185307179586 * u2);
+ }
+};
+
+// One static seed: the window the accelerometer reads at R_true with bias b_a,
+// seeded the way the backend seeds it. Empty when the static gate rejects the
+// window (|f| more than 5% from g).
+std::optional
+try_seed_static(const SO3& R_true, const Vec3& ba, Gauss& n, bs::InitDiagnostics& diag, T g_local = kG) {
+ const T s = kDensity * std::sqrt(kRate); // per-sample sigma
+ const Vec3 f = R_true.inverse() * Vec3{{0, 0, g_local}} + ba;
+ std::vector gyro(kN, Vec3{}), accel(kN);
+ for (auto& a : accel)
+ a = f + Vec3{{s * n(), s * n(), s * n()}};
+ bs::ImuInitConfig cfg;
+ cfg.gravity_magnitude = kG;
+ const bs::ImuInitializer init(cfg);
+ const auto r = init.try_static(std::span{gyro}, std::span{accel});
+ if (!r.success)
+ return std::nullopt;
+ State st(T{1} / T{10});
+ const auto prior = s1::tilt_bias_prior(
+ std::span{accel}, bs::InitMethod::Static, kSigmaBa, kDensity, static_cast(kN) / kRate);
+ s1::seed_from_imu(st, r, bs::InitMethod::Static, 0.0, std::span{accel}, kG, diag, prior);
+ return st;
+}
+
+State seed_static(const SO3& R_true, const Vec3& ba, Gauss& n, bs::InitDiagnostics& diag) {
+ auto st = try_seed_static(R_true, ba, n, diag);
+ REQUIRE(st.has_value());
+ return *st;
+}
+
+// e^T P^-1 e over the tilt (two components orthogonal to the seed's up) and the
+// three accelerometer-bias components. The yaw about up is a gauge, and the
+// truth's yaw is not the seed's, so the tilt error comes from the two up
+// vectors: R = R_hat Exp(d) takes the seed's up u_hat to the truth's u when
+// d_perp = -(u_hat x u) phi / sin(phi), phi the angle between them.
+double tilt_bias_nees(const State& st, const SO3& R_true, const Vec3& ba_true) {
+ const auto P = st.covariance();
+ const Vec3 u = st.R.inverse() * Vec3{{0, 0, 1}};
+ Vec3 a = ld::cross(u, Vec3{{1, 0, 0}});
+ if (ld::norm(a) < 0.1)
+ a = ld::cross(u, Vec3{{0, 1, 0}});
+ a = a * (1.0 / ld::norm(a));
+ const Vec3 b = ld::cross(u, a);
+ // Basis rows: tilt along a, b; bias x, y, z.
+ std::array, 5> B{};
+ for (std::size_t i = 0; i < 3; ++i) {
+ B[0][State::kTheta + i] = a[i];
+ B[1][State::kTheta + i] = b[i];
+ B[2 + i][State::kBa + i] = 1.0;
+ }
+ const Vec3 ut = R_true.inverse() * Vec3{{0, 0, 1}};
+ const Vec3 c = ld::cross(u, ut);
+ const double sphi = ld::norm(c);
+ const double phi = std::atan2(sphi, ld::dot(u, ut));
+ const Vec3 th = sphi > 0 ? c * (-phi / sphi) : Vec3{};
+ const Vec3 db = ba_true - st.ba;
+ const std::array e{ld::dot(a, th), ld::dot(b, th), db[0], db[1], db[2]};
+ std::array, 5> S{};
+ for (std::size_t i = 0; i < 5; ++i)
+ for (std::size_t j = 0; j < 5; ++j)
+ for (std::size_t k = 0; k < 15; ++k)
+ for (std::size_t l = 0; l < 15; ++l)
+ S[i][j] += B[i][k] * P(k, l) * B[j][l];
+ // Solve S y = e by Gaussian elimination (S is SPD, 5 x 5).
+ std::array y = e;
+ for (std::size_t c = 0; c < 5; ++c) {
+ for (std::size_t r = c + 1; r < 5; ++r) {
+ const double m = S[r][c] / S[c][c];
+ for (std::size_t k = c; k < 5; ++k)
+ S[r][k] -= m * S[c][k];
+ y[r] -= m * y[c];
+ }
+ }
+ for (std::size_t c = 5; c-- > 0;) {
+ for (std::size_t k = c + 1; k < 5; ++k)
+ y[c] -= S[c][k] * y[k];
+ y[c] /= S[c][c];
+ }
+ double q = 0;
+ for (std::size_t i = 0; i < 5; ++i)
+ q += e[i] * y[i];
+ return q;
+}
+
+} // namespace
+
+TEST_CASE("S1 prior: P0 is the closed-form tilt-bias block about the seed's up", "[sdk][s1][init][609]") {
+ Gauss n(7);
+ const SO3 R_true = SO3::exp(Vec3{{0.3, -0.2, 0.9}});
+ bs::InitDiagnostics diag;
+ const State st = seed_static(R_true, Vec3{{0.05, 0.1, -0.03}}, n, diag);
+ const auto P = st.covariance();
+ const Vec3 u = st.R.inverse() * Vec3{{0, 0, 1}};
+ const auto ux = ld::hat(u);
+ const T sb2 = kSigmaBa * kSigmaBa;
+ for (std::size_t i = 0; i < 3; ++i) {
+ for (std::size_t j = 0; j < 3; ++j) {
+ INFO("i = " << i << ", j = " << j);
+ // P_theta_b = sigma_ba^2 / g [u]x, and its transpose.
+ CHECK(std::abs(P(State::kTheta + i, State::kBa + j) - sb2 / kG * ux(i, j)) < 1e-12);
+ CHECK(std::abs(P(State::kBa + j, State::kTheta + i) - sb2 / kG * ux(i, j)) < 1e-12);
+ // Nothing else correlates with the tilt: position, velocity, gyro bias.
+ for (const std::size_t k : {State::kPos, State::kVel, State::kBg})
+ CHECK(P(State::kTheta + i, k + j) == 0.0);
+ }
+ }
+ // The yaw about up keeps the fresh state's variance; the tilt is sigma_ba / g.
+ T yaw2 = 0, tilt_trace = 0;
+ for (std::size_t i = 0; i < 3; ++i) {
+ tilt_trace += P(State::kTheta + i, State::kTheta + i);
+ for (std::size_t j = 0; j < 3; ++j)
+ yaw2 += u[i] * P(State::kTheta + i, State::kTheta + j) * u[j];
+ }
+ CHECK(std::abs(yaw2 - 0.01) < 1e-12);
+ const T tilt2 = (tilt_trace - yaw2) / 2;
+ CHECK(tilt2 > sb2 / (kG * kG));
+ CHECK(tilt2 < 1.01 * sb2 / (kG * kG)); // the window's own noise adds under 1%
+ // The bias along up is seeded from |f| - g, and the window resolves it to the
+ // spread of surface gravity, not to sigma_ba.
+ T par2 = 0;
+ for (std::size_t i = 0; i < 3; ++i)
+ for (std::size_t j = 0; j < 3; ++j)
+ par2 += u[i] * P(State::kBa + i, State::kBa + j) * u[j];
+ CHECK(std::sqrt(par2) < 0.03);
+ CHECK(std::sqrt(par2) >= s1::kGravityMagnitudeSigma);
+}
+
+TEST_CASE("S1 prior: the square-root covariance carries the same block", "[sdk][s1][init][609]") {
+ // apply_tilt_bias_prior maps P through G = L_M L_cur^-1 with cov.transform(G).
+ // The square-root backend's transform is S -> G S, retriangularized: the
+ // block it carries must be the dense one, entry by entry.
+ Gauss n(11);
+ const SO3 R_true = SO3::exp(Vec3{{-0.4, 0.25, 2.1}});
+ const T s = kDensity * std::sqrt(kRate);
+ const Vec3 f = R_true.inverse() * Vec3{{0, 0, kG}} + Vec3{{0.08, -0.12, 0.04}};
+ std::vector gyro(kN, Vec3{}), accel(kN);
+ for (auto& a : accel)
+ a = f + Vec3{{s * n(), s * n(), s * n()}};
+ bs::ImuInitConfig cfg;
+ cfg.gravity_magnitude = kG;
+ const auto r = bs::ImuInitializer(cfg).try_static(std::span{gyro}, std::span{accel});
+ REQUIRE(r.success);
+ const auto prior = s1::tilt_bias_prior(
+ std::span{accel}, bs::InitMethod::Static, kSigmaBa, kDensity, static_cast(kN) / kRate);
+ REQUIRE(prior.accel_bias_sigma == kSigmaBa);
+
+ State full(T{1} / T{10});
+ ms::State> sq(T{1} / T{10});
+ bs::InitDiagnostics d1, d2;
+ s1::seed_from_imu(full, r, bs::InitMethod::Static, 0.0, std::span{accel}, kG, d1, prior);
+ s1::seed_from_imu(sq, r, bs::InitMethod::Static, 0.0, std::span{accel}, kG, d2, prior);
+
+ const auto Pf = full.covariance();
+ const auto Ps = sq.covariance();
+ REQUIRE(Ps.rows == Pf.rows);
+ for (std::size_t i = 0; i < Pf.rows; ++i)
+ for (std::size_t j = 0; j < Pf.cols; ++j) {
+ INFO("i = " << i << ", j = " << j);
+ CHECK(std::abs(Ps(i, j) - Pf(i, j)) < 1e-12);
+ }
+ // And it is the block, not the isotropic prior: the tilt-bias correlation is there.
+ CHECK(std::abs(Ps(State::kTheta, State::kBa + 1)) + std::abs(Ps(State::kTheta, State::kBa + 2)) > 1e-4);
+}
+
+TEST_CASE("S1 prior: only a static window sets it", "[sdk][s1][init][609]") {
+ // A gravity-aligned window moves and a dynamic seed's tilt comes from the
+ // alignment: both keep the fresh state's isotropic prior. So does a disabled
+ // prior (sigma_ba = 0, the P0 before #609).
+ const std::vector accel(kN, Vec3{{0.1, 0.2, kG}});
+ const std::span a{accel};
+ const T w = static_cast(kN) / kRate;
+ CHECK(s1::tilt_bias_prior(a, bs::InitMethod::Static, kSigmaBa, kDensity, w).accel_bias_sigma == kSigmaBa);
+ for (const auto m : {bs::InitMethod::GravityAlign, bs::InitMethod::Dynamic, bs::InitMethod::Identity})
+ CHECK(s1::tilt_bias_prior(a, m, kSigmaBa, kDensity, w).accel_bias_sigma == 0.0);
+ CHECK(s1::tilt_bias_prior(a, bs::InitMethod::Static, 0.0, kDensity, w).accel_bias_sigma == 0.0);
+ State st(T{1} / T{10});
+ const auto before = st.covariance().d;
+ CHECK_FALSE(s1::apply_tilt_bias_prior(st, kG, s1::TiltBiasPrior{}));
+ CHECK(st.covariance().d == before);
+}
+
+TEST_CASE("S1 prior: P0 covers a static seed's tilt and bias error (NEES 5)", "[sdk][s1][init][609]") {
+ // The truth's bias is drawn from the prior; each seed's 5-dof error (tilt
+ // orthogonal to up, the three bias components) must have NEES 5 on average.
+ // The static gate bounds the bias along up (|f| within 5% of g), which
+ // truncates that component's draws; the window resolves it to 0.026 m/s^2
+ // whatever its value, so the truncation does not move the NEES.
+ Gauss n(11);
+ constexpr int kDraws = 600;
+ double sum = 0;
+ for (int k = 0; k < kDraws; ++k) {
+ const SO3 R_true = SO3::exp(Vec3{{0.4 * n(), 0.4 * n(), 3.0 * n()}});
+ bs::InitDiagnostics diag;
+ std::optional st;
+ Vec3 ba{};
+ // The site's gravity is not the configured 9.81: it is drawn from the
+ // spread the prior assumes for the bias along up.
+ const T g_local = kG + s1::kGravityMagnitudeSigma * n();
+ while (!st) { // a draw the static gate rejects is drawn again
+ ba = Vec3{{kSigmaBa * n(), kSigmaBa * n(), kSigmaBa * n()}};
+ st = try_seed_static(R_true, ba, n, diag, g_local);
+ }
+ sum += tilt_bias_nees(*st, R_true, ba);
+ }
+ const double anees = sum / kDraws;
+ INFO("ANEES " << anees << " (target 5; the mean of 600 draws has sigma 0.13)");
+ CHECK(anees > 4.5);
+ CHECK(anees < 5.5);
+}
+
+TEST_CASE("S1 prior: a 0.5 m/s^2 bias leaves the seed tilted, and P0 says how much", "[sdk][s1][init][609]") {
+ Gauss n(3);
+ const SO3 R_true = SO3::exp(Vec3{{0.05, -0.1, 0.7}});
+ const Vec3 ba{{0.0, 0.5, 0.0}};
+ bs::InitDiagnostics diag;
+ const State st = seed_static(R_true, ba, n, diag);
+ const Vec3 u_true = R_true.inverse() * Vec3{{0, 0, 1}};
+ const Vec3 u_seed = st.R.inverse() * Vec3{{0, 0, 1}};
+ const double tilt = std::atan2(ld::norm(ld::cross(u_true, u_seed)), ld::dot(u_true, u_seed));
+ const auto P = st.covariance();
+ T tilt_trace = 0, yaw2 = 0;
+ for (std::size_t i = 0; i < 3; ++i) {
+ tilt_trace += P(State::kTheta + i, State::kTheta + i);
+ for (std::size_t j = 0; j < 3; ++j)
+ yaw2 += u_seed[i] * P(State::kTheta + i, State::kTheta + j) * u_seed[j];
+ }
+ const double tilt_sigma = std::sqrt((tilt_trace - yaw2) / 2);
+ INFO("seed tilt " << tilt * 57.29578 << " deg, P0 tilt sigma " << tilt_sigma * 57.29578 << " deg");
+ CHECK(tilt > 0.04); // about 0.5 / g: the bias is in the tilt
+ CHECK(tilt < 3.0 * tilt_sigma); // and P0 covers it
+ CHECK(tilt_bias_nees(st, R_true, ba) < 25.0); // 2.5 sigma on 5 dof, from one draw
+}
+
+TEST_CASE("MSCKF backend: a 0.5 m/s^2 accelerometer bias is learned once the motion rotates",
+ "[sdk][msckf][backend][609]") {
+ // The synthetic world starts at rest (static init fires) and then loops with
+ // an oscillating yaw. Its accelerometer bias is 0.5 m/s^2 on y.
+ namespace ev = branes::sdk::eval;
+ ev::SyntheticConfig cfg;
+ cfg.accel_bias = Vec3{{0.0, 0.5, 0.0}};
+ cfg.duration_s = 30.0;
+ const auto world = ev::generate_world(cfg);
+
+ const auto run = [&](double prior_sigma) {
+ using Backend = bs::MsckfBackend;
+ Backend::CameraCalibration cal;
+ cal.intrinsics = world.camera;
+ cal.extrinsics.R_imu_cam = world.R_imu_cam;
+ cal.extrinsics.p_imu_cam = world.p_imu_cam;
+ Backend backend(std::vector{cal});
+ bs::VioConfig vcfg;
+ vcfg.accel_bias_prior_sigma = prior_sigma;
+ backend.initialize(vcfg);
+ struct Out {
+ double tilt_first = -1, sigma_first = 0, tilt_last = 0, ba_y = 0;
+ } o;
+ std::size_t imu_idx = 0;
+ for (std::size_t f = 0; f < world.frames.size(); ++f) {
+ const double t = world.frames[f].t;
+ const Vec3 u_true = world.gt[f].R.inverse() * Vec3{{0, 0, 1}};
+ for (; imu_idx < world.imu.size() && world.imu[imu_idx].timestamp_s <= t; ++imu_idx) {
+ backend.process_imu(world.imu[imu_idx]);
+ // The seed itself, before propagation grows P (the start is at
+ // rest, so the truth's attitude is the same at every warm-up time).
+ if (backend.initialized() && o.tilt_first < 0) {
+ const auto& st = backend.state();
+ const Vec3 u = st.R.inverse() * Vec3{{0, 0, 1}};
+ const auto P = st.covariance();
+ T tr = 0, yaw2 = 0;
+ for (std::size_t i = 0; i < 3; ++i) {
+ tr += P(State::kTheta + i, State::kTheta + i);
+ for (std::size_t j = 0; j < 3; ++j)
+ yaw2 += u[i] * P(State::kTheta + i, State::kTheta + j) * u[j];
+ }
+ o.tilt_first = std::atan2(ld::norm(ld::cross(u_true, u)), ld::dot(u_true, u));
+ o.sigma_first = std::sqrt((tr - yaw2) / 2);
+ }
+ }
+ backend.process_camera(t, std::span>{world.frames[f].obs});
+ if (!backend.initialized())
+ continue;
+ const auto& st = backend.state();
+ const Vec3 u_est = st.R.inverse() * Vec3{{0, 0, 1}};
+ o.tilt_last = std::atan2(ld::norm(ld::cross(u_true, u_est)), ld::dot(u_true, u_est));
+ o.ba_y = st.ba[1];
+ }
+ return o;
+ };
+ const auto shipped = run(0.2);
+ const auto isotropic = run(0.0);
+ constexpr double kDeg = 57.29577951308232;
+ INFO("window-derived P0: tilt " << shipped.tilt_first * kDeg << " deg (sigma " << shipped.sigma_first * kDeg
+ << ") -> " << shipped.tilt_last * kDeg << " deg, b_a y " << shipped.ba_y);
+ INFO("isotropic P0: tilt " << isotropic.tilt_first * kDeg << " deg (sigma " << isotropic.sigma_first * kDeg
+ << ") -> " << isotropic.tilt_last * kDeg << " deg, b_a y " << isotropic.ba_y);
+ REQUIRE(world.frames.front().t >= static_cast(cfg.warmup_s) / 2); // the seed is at rest
+ CHECK(shipped.tilt_first > 0.04); // the seed takes the bias as tilt
+ CHECK(shipped.tilt_first < 3.0 * shipped.sigma_first); // and its P0 covers it
+ CHECK(shipped.sigma_first < 0.03); // with sigma_ba / g, about 1.2 deg
+ CHECK(shipped.tilt_last < 0.5 / kDeg); // the motion separates them
+ CHECK(std::abs(shipped.ba_y - 0.5) < 0.1); // and the bias is learned
+ CHECK(shipped.tilt_last < isotropic.tilt_last);
+}
diff --git a/tests/tools/invariant_backend_adapter.cpp b/tests/tools/invariant_backend_adapter.cpp
index 5bf0456..9dc1d6c 100644
--- a/tests/tools/invariant_backend_adapter.cpp
+++ b/tests/tools/invariant_backend_adapter.cpp
@@ -246,6 +246,45 @@ TEST_CASE("invariant adapter: initializes through the shipped MSCKF with its see
CHECK(max_abs_diff(a.covariance(), expected) < 1e-15);
}
+TEST_CASE("invariant adapter: the seed's tilt-bias correlation reaches the invariant prior",
+ "[tools][invariant_adapter][609]") {
+ // Since #609 the shipped seed correlates the body tilt with the accelerometer
+ // bias. The invariant error's attitude is the world one, R_hat times the body
+ // one, and b_a is unchanged, so the cross block must arrive as R_hat P_theta_b.
+ namespace ms = branes::sdk::msckf;
+ using Boot = bs::MsckfBackend;
+ using St = ms::State;
+ Adapter a = make_adapter();
+ Boot boot(std::vector{
+ Boot::CameraCalibration{euroc_like_camera(), ms::CameraExtrinsics{SO3{}, Vec3{}}}});
+ const bs::VioConfig cfg;
+ a.initialize(cfg);
+ boot.initialize(cfg);
+ for (int i = 0; i < 400 && !(boot.initialized() && a.initialized()); ++i) {
+ const auto m = tilted_sample(0.005 * i);
+ if (!boot.initialized())
+ boot.process_imu(m);
+ if (!a.initialized())
+ a.process_imu(m);
+ }
+ REQUIRE(boot.initialized());
+ REQUIRE(a.initialized());
+ const auto Pb = boot.state().covariance();
+ const auto Pi = a.covariance();
+ const auto R = boot.state().R.matrix();
+ double cross = 0;
+ for (std::size_t i = 0; i < 3; ++i) {
+ for (std::size_t j = 0; j < 3; ++j) {
+ double e = 0;
+ for (std::size_t k = 0; k < 3; ++k)
+ e += R(i, k) * Pb(St::kTheta + k, St::kBa + j);
+ CHECK(std::abs(Pi(ms::InvariantNavState::kTheta + i, ms::InvariantNavState::kBa + j) - e) < 1e-15);
+ cross = std::max(cross, std::abs(e));
+ }
+ }
+ CHECK(cross > 1e-3); // the correlation is there: sigma_ba^2 / g = 4e-3
+}
+
TEST_CASE("invariant adapter: S8 holds a frame at rest out of the window", "[tools][invariant_adapter]") {
const auto run = [](Adapter::Options opt) {
Adapter a = make_adapter(opt);
diff --git a/tools/include/branes/tools/bench/s1_initialization_bench.hpp b/tools/include/branes/tools/bench/s1_initialization_bench.hpp
index d0bc369..932ce42 100644
--- a/tools/include/branes/tools/bench/s1_initialization_bench.hpp
+++ b/tools/include/branes/tools/bench/s1_initialization_bench.hpp
@@ -27,6 +27,12 @@
// sfm_adjacent_bootstrap
// the same with the bootstrap on frame 1 (the pre-#247 choice)
//
+// The static and gravity-aligned variants seed the backend's mean with the
+// fresh state's isotropic P₀. The backend then replaces the tilt and
+// accelerometer-bias blocks with the prior its window implies (#609,
+// s1_initialization::tilt_bias_prior), which needs the window's duration that
+// these inputs do not carry; tests/sdk/tilt_bias_prior.cpp checks that block.
+//
// Fixtures:
// known_answer a noiseless, tilted, gyro-biased static window: the expected
// seed (attitude, biases, P₀, diagnostics) in closed form
@@ -118,7 +124,8 @@ struct S1InitializationBench {
}
[[nodiscard]] static std::vector variants() {
- return {{"shipped", "try_static -> seed_from_imu (backend default)"},
+ return {{"shipped",
+ "try_static -> seed_from_imu (the backend's static seed, with the isotropic P0: no tilt-bias prior)"},
{"gravity_align", "try_gravity_align -> seed_from_imu (moving-start fallback)"},
{"dynamic", "try_dynamic -> seed_from_alignment (visual-inertial alignment)"},
{"sfm_two_view_pnp",
diff --git a/tools/include/branes/tools/pose_consistency.hpp b/tools/include/branes/tools/pose_consistency.hpp
index 0238332..5600f09 100644
--- a/tools/include/branes/tools/pose_consistency.hpp
+++ b/tools/include/branes/tools/pose_consistency.hpp
@@ -67,6 +67,15 @@ namespace branes::tools {
/// Which error-state convention the covariance is written in.
enum class PoseErrorConvention { body_right, right_invariant };
+/// The last scored frame's roll/pitch: what the tilt probe (#609) records per
+/// frame. The error and σ are the world roll/pitch components (rad); the NEES
+/// is their 2-dof quadratic form.
+struct TiltFrame {
+ double err_rad = 0.0; ///< |world roll/pitch error|
+ double sigma_rad = 0.0; ///< √(trace of the 2 × 2 roll/pitch covariance / 2)
+ double nees = 0.0; ///< eᵀ P⁻¹ e over roll/pitch, target 2
+};
+
/// Per-run summary: ANEES by block, and the world-yaw σ over the run.
struct PoseConsistencySummary {
std::size_t frames = 0; ///< frames scored (a ground-truth match and a PD covariance)
@@ -239,6 +248,7 @@ class PoseConsistency {
nees_[1] += *rp;
nees_[2] += *yaw;
nees_[3] += *ps;
+ last_tilt_ = TiltFrame{std::hypot(e[0], e[1]), std::sqrt(0.5 * (Pw[0][0] + Pw[1][1])), *rp};
const double ys = std::sqrt(Pw[2][2]);
const double ye = std::abs(e[2]);
@@ -266,6 +276,11 @@ class PoseConsistency {
return s;
}
+ /// Roll/pitch of the last frame `add` scored (zeros before the first).
+ [[nodiscard]] const TiltFrame& last_tilt() const {
+ return last_tilt_;
+ }
+
/// The truth→estimate gauge fixed at the first scored frame, if any.
[[nodiscard]] const std::optional& anchor() const {
return anchor_;
@@ -296,6 +311,7 @@ class PoseConsistency {
std::optional anchor_;
std::array nees_{};
PoseConsistencySummary sum_;
+ TiltFrame last_tilt_;
};
} // namespace branes::tools
diff --git a/tools/src/vio_pipeline.cpp b/tools/src/vio_pipeline.cpp
index 4fdddb6..bbd1be7 100644
--- a/tools/src/vio_pipeline.cpp
+++ b/tools/src/vio_pipeline.cpp
@@ -115,9 +115,10 @@ struct Args {
bool prefer_dynamic = false; ///< EuRoC: dynamic VI init only (VioConfig::prefer_dynamic_init)
double gate_alpha = VioConfig{}.gate_alpha; ///< MSCKF S6d chi-square gate significance (#481)
bool gate_per_dof = false; ///< MSCKF S6d: the pre-#481 5-per-dof gate
- bool gate_flag = false; ///< --gate-alpha or --gate-per-dof was given
- bool detect_trackable_only = false; ///< S4: skip the KLT border band when detecting (#474)
- bool observability_ledger = false; ///< per-stage spurious yaw/position information (#480)
+ double accel_bias_prior_sigma = VioConfig{}.accel_bias_prior_sigma; ///< S1 seed prior (#609); 0 = isotropic
+ bool gate_flag = false; ///< --gate-alpha or --gate-per-dof was given
+ bool detect_trackable_only = false; ///< S4: skip the KLT border band when detecting (#474)
+ bool observability_ledger = false; ///< per-stage spurious yaw/position information (#480)
// Composition capture (#447): the loop over post-init frames [first, last].
std::optional> capture_loop;
// Ideal tracks (#602): --tracks gt replaces the KLT front end with
@@ -145,6 +146,7 @@ VioConfig config_of(const Args& a) {
VioConfig cfg;
cfg.gate_alpha = a.gate_alpha;
cfg.gate_per_dof = a.gate_per_dof;
+ cfg.accel_bias_prior_sigma = a.accel_bias_prior_sigma;
if (a.camera_sigma_px > 0.0) // in normalized units through cam0's focal length
cfg.camera_noise_normalized = a.camera_sigma_px / euroc_cam0().intrinsics.fx();
return cfg;
@@ -279,6 +281,12 @@ Args parse(int argc, char** argv) {
a.gate_flag = true;
if (!(a.gate_alpha > 0.0 && 1.0 - a.gate_alpha < 1.0 && 1.0 - a.gate_alpha > 0.0))
throw std::invalid_argument("--gate-alpha needs a significance in (0,1)");
+ } else if (v == "--accel-bias-prior-sigma") {
+ if (!next_is_value())
+ throw std::invalid_argument("--accel-bias-prior-sigma needs a sigma in m/s^2 (0 = isotropic)");
+ a.accel_bias_prior_sigma = std::stod(argv[++i]);
+ if (!(a.accel_bias_prior_sigma >= 0.0))
+ throw std::invalid_argument("--accel-bias-prior-sigma needs a sigma >= 0");
} else if (v == "--detect-trackable-only")
a.detect_trackable_only = true;
else if (v == "--riekf-legacy") {
@@ -821,6 +829,60 @@ void write_tum(const Args& args, const std::vector>& traj) {
branes::tools::write_tum(f, traj);
}
+// /tilt.csv (#609): per scored frame, the world roll/pitch error against
+// the ground truth, the filter's roll/pitch σ and NEES (from PoseConsistency),
+// and its accelerometer bias with σ next to the ground truth's
+// (state_groundtruth_estimate0; NaN with --imu gt, whose bias is synthesized).
+class TiltLog {
+public:
+ explicit TiltLog(const Args& args) {
+ if (args.out.empty())
+ return;
+ f_.open(args.out + "/tilt.csv");
+ if (!f_) {
+ std::cerr << "vio_pipeline: cannot write " << args.out << "/tilt.csv\n";
+ return;
+ }
+ f_ << std::setprecision(9)
+ << "t,tilt_err_deg,tilt_sigma_deg,tilt_nees,ba_x,ba_y,ba_z,ba_sx,ba_sy,ba_sz,gt_ba_x,gt_ba_y,gt_ba_z\n";
+ if (!args.gt_imu) {
+ try {
+ gt_ = euroc::parse_groundtruth_states(args.dataset);
+ } catch (const std::exception&) {
+ gt_.clear(); // no bias column: the GT bias reads NaN
+ }
+ }
+ }
+ void add(double t, const branes::tools::TiltFrame& tf, const Vec3& ba, const msckf::DynMat& P, std::size_t kba) {
+ if (!f_.is_open())
+ return;
+ constexpr double kDeg = 180.0 / std::numbers::pi;
+ f_ << std::fixed << std::setprecision(6) << t << std::defaultfloat << std::setprecision(9) << ','
+ << tf.err_rad * kDeg << ',' << tf.sigma_rad * kDeg << ',' << tf.nees;
+ for (std::size_t i = 0; i < 3; ++i)
+ f_ << ',' << static_cast(ba[i]);
+ for (std::size_t i = 0; i < 3; ++i)
+ f_ << ',' << std::sqrt(static_cast(P(kba + i, kba + i)));
+ const auto g = gt_ba(t);
+ for (std::size_t i = 0; i < 3; ++i)
+ f_ << ',' << g[i];
+ f_ << '\n';
+ }
+
+private:
+ [[nodiscard]] std::array gt_ba(double t) const {
+ constexpr double nan = std::numeric_limits::quiet_NaN();
+ const auto it = std::lower_bound(
+ gt_.begin(), gt_.end(), t, [](const euroc::GroundTruthState& g, double x) { return g.t_s < x; });
+ if (it == gt_.end() || std::abs(it->t_s - t) > 0.02)
+ return {nan, nan, nan};
+ return {
+ static_cast(it->nav.ba[0]), static_cast(it->nav.ba[1]), static_cast(it->nav.ba[2])};
+ }
+ std::ofstream f_;
+ std::vector> gt_;
+};
+
// --imu gt (#613): replace the real IMU with one synthesized from a smooth truth
// through the ground truth, at the real IMU's timestamps and the filter's own
// noise densities, and make that truth the ground truth for the ideal tracks,
@@ -942,6 +1004,7 @@ bool run_euroc(const Args& args,
ideal.emplace(gt, cam0.intrinsics, branes::math::lie::SE3(cam0.R_imu_cam, cam0.p_imu_cam), args.ideal);
std::size_t ideal_obs = 0;
branes::tools::PoseConsistency consistency(gt);
+ TiltLog tilt(args);
std::vector> traj;
std::size_t imu_idx = 0, feat_total = 0, n = 0;
@@ -976,12 +1039,17 @@ bool run_euroc(const Args& args,
traj.push_back(sp);
++n;
sigma_sum += pos_sigma(est.backend().state().covariance(), msckf::State::kPos);
- consistency.add(t,
- sp.pose,
- est.backend().state().covariance(),
- msckf::State::kTheta,
- msckf::State::kPos,
- branes::tools::PoseErrorConvention::body_right);
+ if (consistency.add(t,
+ sp.pose,
+ est.backend().state().covariance(),
+ msckf::State::kTheta,
+ msckf::State::kPos,
+ branes::tools::PoseErrorConvention::body_right))
+ tilt.add(t,
+ consistency.last_tilt(),
+ est.backend().state().ba,
+ est.backend().state().covariance(),
+ msckf::State::kBa);
const auto tf = est.tracked_features();
feat_total += ideal ? ideal_obs : tf.size();
const double nis = safe_nis(est.backend().nis_consistency());
@@ -1028,7 +1096,7 @@ bool run_euroc(const Args& args,
bool run_euroc_invariant(const Args& args, const VioConfig& cfg, RunResult& out, std::ofstream* stream) {
using Adapter = branes::tools::InvariantBackendAdapter;
using Estimator = branes::sdk::VioEstimator;
- constexpr std::size_t kInvTheta = 0, kInvPos = 6; // invariant nav error-state: δθ(0) δv(3) δp(6)
+ constexpr std::size_t kInvTheta = 0, kInvPos = 6, kInvBa = 12; // invariant nav error: δθ δv δp δbg δba
std::vector> imu;
std::vector images;
@@ -1060,6 +1128,7 @@ bool run_euroc_invariant(const Args& args, const VioConfig& cfg, RunResult& out,
ideal.emplace(gt, cam0.intrinsics, branes::math::lie::SE3(cam0.R_imu_cam, cam0.p_imu_cam), args.ideal);
std::size_t ideal_obs = 0;
branes::tools::PoseConsistency consistency(gt);
+ TiltLog tilt(args);
std::vector> traj;
std::size_t imu_idx = 0, feat_total = 0, n = 0;
@@ -1095,7 +1164,8 @@ bool run_euroc_invariant(const Args& args, const VioConfig& cfg, RunResult& out,
++n;
const auto P = est.backend().covariance();
sigma_sum += pos_sigma(P, kInvPos);
- consistency.add(t, sp.pose, P, kInvTheta, kInvPos, branes::tools::PoseErrorConvention::right_invariant);
+ if (consistency.add(t, sp.pose, P, kInvTheta, kInvPos, branes::tools::PoseErrorConvention::right_invariant))
+ tilt.add(t, consistency.last_tilt(), est.backend().impl().nav().ba, P, kInvBa);
feat_total += ideal ? ideal_obs : est.num_tracked_features();
if (stream && stream->is_open()) {
const Vec3 ep = est.current_pose().translation();
@@ -1199,6 +1269,8 @@ int main(int argc, char** argv) {
" with --capture-stages S1_initialization, captures every attempt\n"
" --gate-alpha A MSCKF: reject a track above the chi-square(dof) quantile at\n"
" significance A (default 0.05)\n"
+ " --accel-bias-prior-sigma S S1: the seed's accelerometer-bias prior sigma, m/s^2 (default\n"
+ " 0.2); 0 restores the isotropic prior before #609\n"
" --gate-per-dof MSCKF: gate at 5 per dof instead (the gate before #481)\n"
" (both also select the R-IEKF's gate on EuRoC, #608)\n"
" --riekf-legacy LIST R-IEKF on EuRoC: run these pieces as before #608, to measure\n"