D&A decouples description and assessment for zero-shot robotic reward models
Original titleD&A: Decoupling Description and Assessment Enables VLMs as Zero-Shot Robotic Reward Models
- Outlet
- Takara TLDR
- Published
- Importance
- 2/5
Summary
Researchers present D&A, a method that separates progress estimation into scene description and assessment phases to mitigate confirmation bias in vision-language models used as robotic reward models. The prompt-based approach generalizes across tasks and embodiments while enabling efficient downstream reinforcement learning.
Original excerpt
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.