robotics
The Planner Could Only Emit an Action the World Model Allowed
Visual grounding built a symbolic state, then constrained decoding and search produced executable household plans under that model.
Summary
Visual grounding built a symbolic state, then constrained decoding and search produced executable household plans under that model.
The agent first explores with a vision-language model to acquire goal-relevant predicates and object bindings, then uses a PDDL transition model to restrict decoding to applicable actions while Monte Carlo tree search evaluates continuations. Open models from 4B to 27B parameters exceeded 90 percent reported success on VirtualHome and ALFWorld. In ALFWorld, constraints or search alone solved under one-third of tasks, while their combination exceeded 95 percent; residual failures moved toward visual state acquisition, and transfer still depends on correct grounding.
Why it matters
Visual grounding built a symbolic state, then constrained decoding and search produced executable household plans under that model.
Limits and context
No additional limitation was separately recorded.
Key claims
Visual grounding built a symbolic state, then constrained decoding and search produced executable household plans under that model.
Evidence: source-2026-08-18-009
Sources
- arXiv preprint 2608.16794arXiv · primary research
Corrections
No corrections have been recorded for this story.