D&A, VLM 기반 로봇 보상 모델의 편향 제거 기법 제안
D&A: Decoupling Description and Assessment Enables VLMs as Zero-Shot Robotic Reward Models
- 매체
- Takara TLDR
- 게시
- 중요도
- 2/5
- 화제 지수
- —
한국어 요약
시각-언어 모델(VLM)을 제로샷 로봇 보상 모델로 활용할 때 발생하는 확인 편향을 해결하는 D&A 기법이 제안되었다. 장면 설명과 평가 단계를 분리해 명시적 기준으로 평가를 고정함으로써 세밀한 진행률 추정과 실패 감지가 가능하며, 작업과 구체화에 걸쳐 일반화된다.
원문 발췌 · English
Vision-Language Models (VLMs) hold promise for zero-shot video task-progress estimation in robot learning, yet existing frameworks frequently suffer from confirmation bias, hallucinating progress in failed executions due to implicit success assumptions or relative temporal ranking designs. Naive prompting of frontier models collapses into a binary success detector that fails to capture nuanced, non-linear progress and failures. While post-training approaches can mitigate this through explicit training, they demand expensive real-world robotics data and require computationally intensive post-training for every individual base model. To resolve these limitations, we introduce D&A (Describe & Assess): our approach decouples progress estimation into an objective, task-grounded textual description of scene dynamics followed by an assessment phase using explicit rubrics. By compelling the VLM to synthesize an explicit, text-based definition of absolute success before scoring, the framework anchors evaluations to an external, global criterion rather than the video's internal timeline. Empirical results demonstrate that our method calibrates zero-shot estimates and mitigates confirmation bias, providing fine-grained, partial progress metrics that accurately penalize failed executions while correctly interpolating successful runs. Being purely prompt-based, D&A generalizes across tasks, embodiments, and benchmarks---matching or surpassing post-trained reward models---and its reward estimates enable efficient downstream reinforcement learning.