NavBind: Diagnosing Instruction-Action Binding at Route Forks in Navigation VLAs
We introduce NavBind to diagnose instruction-action binding: whether changing the instruction influences the probabilistic distribution of the actions, with the navigation context fixed.
ProofPaper ↗
Key points
- Vision-language-action (VLA) navigation models translate visual observations and instructions into actions.
- Navigation success assesses whether an agent reaches the goal.
- We construct 3,252 route-fork contexts by recombining route-aligned instruction segments from R2R-CE and RxR-CE.
- We compare generated local action sequences, the likelihoods of fixed reference action sequences under both instructions, and changes in visual attention and attribution.
Sources (1)
- [1]NavBind: Diagnosing Instruction-Action Binding at Route Forks in Navigation VLAsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 21, 04:50 AM
We introduce NavBind to diagnose instruction-action binding: whether changing the instruction influences the probabilistic distribution of the actions, with the navigation context fixed.
Vision-language-action (VLA) navigation models translate visual observations and instructions into actions.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 21, 2026From VLM Planners to Robotic Agents: Self-Evolving Harnesses for History-Dependent Manipulation
- Oct 21, 2026Agentic World Action Models
- Oct 20, 2026Perceives but Doesn't Brake: A Perception-Action Binding Gap in Reasoning Driving VLAs
- Oct 20, 2026AgenticRoboRL: Bringing Coding-Agent Post-Training to Agentic Robotics
- Oct 20, 2026RoboCasa-Shift: Stage-Aware Robustness Evaluation of Vision-Language-Action Policies in Long-Horizon Household Manipulation
- Oct 19, 2026Evaluating Agents on Highly Subjective Tasks