Action Hallucination in Generative Visual-Language-Action Models
This addresses reliability issues in robot foundation models for robotics researchers and practitioners, though it is incremental as it builds on existing analysis of generative policies.
The paper analyzed action hallucinations in generative vision-language-action models for robotics, showing these hallucinations arise from structural mismatches between feasible robot behavior and model architectures, and identified three barriers (topological, precision, and horizon) that impose unavoidable tradeoffs.
Robot Foundation Models such as Vision-Language-Action models are rapidly reshaping how robot policies are trained and deployed, replacing hand-designed planners with end-to-end generative action models. While these systems demonstrate impressive generalization, it remains unclear whether they fundamentally resolve the long-standing challenges of robotics. We address this question by analyzing action hallucinations that violate physical constraints and their extension to plan-level failures. Focusing on latent-variable generative policies, we show that hallucinations often arise from structural mismatches between feasible robot behavior and common model architectures. We study three such barriers -- topological, precision, and horizon -- and show how they impose unavoidable tradeoffs. Our analysis provides mechanistic explanations for reported empirical failures of generative robot policies and suggests principled directions for improving reliability and trustworthiness, without abandoning their expressive power.