Sriyash/cube

[yandaboa]

Description

Please include a summary of the change and which issue is fixed. Please also include relevant motivation and context.
List any dependencies that are required for this change.

Fixes # (issue)

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (existing functionality will not work without user modification)
• Documentation update

Screenshots

Please attach before and after screenshots of the change if applicable.

Checklist

• I have read and understood the contribution guidelines
• I have run the pre-commit checks with ./uwlab.sh --format
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• I have updated the changelog and the corresponding version in the extension's config/extension.toml file
• I have added my name to the CONTRIBUTORS.md or my name already exists there

[patrickhaoy]

@greptileai

[greptile-apps]

Greptile Summary

This PR adds cube manipulation configurations, tactile collection and reset-state tooling, diffusion-policy integration, and an ASTEROID iterative collection/training/evaluation workflow.

• Adds new task configurations, observations, rewards, events, actions, and recording support for cube workflows.
• Extends policy evaluation, demonstration collection, and dataset handling.
• Adds a diffusion-policy submodule and orchestration/helper scripts.

Confidence Score: 2/5

The PR is not safe to merge until the broken collection invocation, incompatible shell interpreter, and ignored no-video setting are corrected.

The new orchestration workflow can fail before collecting data because it passes an unsupported argument, the new executable helper fails under its declared shell, and evaluation disregards an explicit resource-control option.

Files Needing Attention: scripts/ASTEROID/run_asteroid.py; scripts_v2/tools/collect_cubepick_expert5.sh

Important Files Changed

Filename	Overview
scripts/ASTEROID/run_asteroid.py	Adds iterative collection, training, and evaluation orchestration, but its collection video flag is incompatible with the target CLI and its no-video option is ignored during evaluation.
scripts_v2/tools/collect_demos.py	Reworks collection around exploration and expert policies, per-environment horizons, recording, and optional video.
scripts_v2/tools/collect_cubepick_expert5.sh	Adds a cube expert collection launcher that cannot execute under its declared POSIX-shell interpreter.
source/uwlab_rl/uwlab_rl/wrappers/diffusion.py	Substantially extends diffusion-policy observation and action adaptation.
source/uwlab_tasks/uwlab_tasks/manager_based/manipulation/omnireset/mdp/events.py	Adds extensive cube reset and state-management behavior.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart LR
  A[ASTEROID orchestrator] --> B[Collect demonstrations]
  B --> C[Train diffusion policy]
  C --> D[Evaluate checkpoint]
  D --> E{More iterations?}
  E -- Yes --> B
  E -- No --> F[Finish]

Loading

_{Reviews (1): Last reviewed commit: "Move ICE driver to scripts/ASTEROID/run_..." | Re-trigger Greptile}

[greptile-apps]

Collection uses unsupported video flag

During a default ASTEROID run, the orchestrator passes --save_video, but collect_demos.py defines only --video, causing collection to fail during argument or Hydra configuration parsing before the first dataset is collected.

[greptile-apps]

Interpreter rejects Bash array syntax

When this executable runs on a system where /bin/sh is a POSIX shell such as dash, its Bash-only arrays and ${!ckpts[@]} expansion fail during parsing, so the collection commands never launch.

Suggested change

	#!/bin/sh
	#!/usr/bin/env bash

[greptile-apps]

No-video setting is ignored

When the orchestrator is run with --no_video, eval_policy still appends --save_video and --enable_cameras, causing evaluation to initialize rendering resources and record video despite the explicit opt-out.