robotics
The Robot Learns From the State Its Mistake Left Behind
A recovery benchmark and online training method reward both progress and verified recovery after manipulation errors.
Summary
A recovery benchmark and online training method reward both progress and verified recovery after manipulation errors.
Execution errors leave embodied agents in states that a clean demonstration never visits. ERR-Env supplies feedback on progress and errors, while ERR-Bench evaluates nominal, intermediate and error states. Training a compact 2-billion-parameter model with action-grouped turn-level reinforcement learning improved strict error-free success by 10.25 percentage points and recovery rate by 22.95 points. The result is benchmark evidence, not proof of open-world recovery.
Why it matters
A recovery benchmark and online training method reward both progress and verified recovery after manipulation errors.
Limits and context
- The result is benchmark evidence, not proof of open-world recovery.
Key claims
A recovery benchmark and online training method reward both progress and verified recovery after manipulation errors.
Qualification: The result is benchmark evidence, not proof of open-world recovery.
Evidence: source-2026-10-21-004
Sources
- arXiv preprint 2610.27003, original v1arXiv · primary research
Corrections
No corrections have been recorded for this story.