Repository navigation
Publish PiPER grip validation, bounded recovery and full confirmation - #27
Merged
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Question
Does a test lift plus bounded recovery improve PiPER acquisition and placement? This extends the physical-grounding research with an opt-in authored guard and a complete paired confirmation.
Evidence
The result is a regression: original placement 4/6 versus guarded 2/6. Both normal cases complete, both intermediate-friction baselines complete while the guard aborts, and neither low-friction case completes. No successful recovery is demonstrated. All 30 full-duration attempts are valid; all 18 negative controls fail. The paper separates probe qualification, short following proxies, alarm motion, retry and completed placement.
The complete byte-identical 36-file original bundle includes all 30 compressed trajectories, the protocol, report, integrity seal and original MIT model notices. The execution wrapper sits outside that seal. Source and the Hugging Face evidence are pinned. The homepage, catalog, feed, README and hardware paper link the new result.
Reproduction
Frozen runtime
3ee94706f436690c6d1d8554eda9202d23620e3c, confirmation seed 731209. Python 3.12.9, MuJoCo 3.15.0 and NumPy 2.5.3 on the recorded CPU host. Installed wheel ran with network denied. Full command:104.261 wall seconds/600 simulated seconds;53/30000 control windows exceeded20ms. Native simulator state is privileged; no learned policy, calibrated sensor/contact model or hardware qualification is claimed.Checks