D&A: Decoupling Description and Assessment Enables VLMs as Zero-Shot Robotic Reward Models
arXiv cs.AI · 3 Nov 2026, 04:52 AEST
Vision-Language Models (VLMs) show potential for zero-shot video task-progress estimation in robot learning. The D&A approach decouples progress estimation into a description and assessment phase, improving accuracy and reducing confirmation bias.
Key facts
- 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.
- 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.
- 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.
Read aloud
Vision-Language Models VLMs show potential for zero-shot video task-progress estimation in robot learning. The D&A approach decouples progress estimation into a description and assessment phase, improving accuracy and reducing confirmation bias. 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. 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. 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.
Sources in this cluster
- D&A: Decoupling Description and Assessment Enables VLMs as Zero-Shot Robotic Reward Models
arXiv cs.AI · 3 Nov 2026, 04:52 AEST