FailBank, 로봇 조작 정책의 런타임 피드백 기반 자체 진화 프레임워크
Learning from Runtime Feedback through Failure-Bank Self-Evolution for Vision-Language-Action Models
- 매체
- Hugging Face
- 게시
- 중요도
- 3/5
- 화제 지수
- 10
한국어 요약
시각-언어-행동(VLA) 모델의 정책을 런타임 피드백으로 지속적으로 개선하는 FailBank 프레임워크가 제안됐다. CBF 기반 안전 모듈의 반사실적 수정을 학습 신호로 변환하여 VLA-Arena 벤치마크에서 작업 성공률을 6.9~8.5% 향상시키고 누적 비용을 23.8~35.6% 감소시켰다.
원문 발췌 · English
Vision-language-action (VLA) models generalize broadly across robotic manipulation tasks, but complex environments require balancing task success with unintended contact. Runtime shields can correct individual actions, but they leave the underlying policy unchanged, so repeated disagreements may create a persistent policy-shield mismatch that blocks task progress. To address this challenge, we introduce FailBank, a four-stage self-evolving framework that converts runtime feedback into persistent policy improvement. During collection, a fixed CBF-based safety module serves as an observe-only teacher, producing counterfactual corrections while the policy remains in control. Outcome-aware admission then converts useful proposals into corrective targets and retains successful uncorrected actions as quiet anchors for guarded LoRA updates. We evaluate FailBank on the VLA-Arena benchmark across two difficulty levels and two VLA backbones. Compared with the base policies, FailBank improves the joint success-cost operating point. Across the two backbones, FailBank improves task success rate by 8.5 and 6.9 percentage points, while reducing policy-induced cumulative cost by 35.6\% and 23.8\%, respectively. Compared with runtime shielding, FailBank raises task success rate by 25.4 and 9.5 percentage points, while maintaining comparable policy-induced cumulative cost. These results show that runtime feedback can serve as persistent policy supervision rather than only as a temporary action constraint.