Generative VLAs hallucinate physically invalid actions due to topological, precision, and horizon mismatches between model architectures and feasible robot behavior.
Embodiedbench: Com- prehensive benchmarking multi-modal large language models for vision-driven embodied agents
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Action Hallucination in Generative Vision-Language-Action Models
Generative VLAs hallucinate physically invalid actions due to topological, precision, and horizon mismatches between model architectures and feasible robot behavior.