Planning Takes More Than Token Prediction: Causal Plan for Benchmarking and Building Physically Grounded Embodied Reasoners

תקציר מקורי באנגליתarXiv:2606.01810v3 Announce Type: replace Abstract: Current benchmarks for embodied vision-language planning inadvertently favor linguistic next-token prediction over physically grounded next-state reasoning. This rewards models that mimic statistical language priors rather than track true causal dependencies, reducing complex physical planning to shallow sequence modeling. Hence, achieving genuine physical autonomy requires a fundamental shift from linguistically grounded token prediction toward physically grounded causal reasoning. To this end, we introduce Causal-Plan-Bench, a high-fidelity diagnostic suite spanning four causal dimensions, curated via multi-stage verification. To endow models with this capability, a four-stage annotation pipeline extracts structured interaction records