Get Started
Home
Topics
Search
Library
Research questionHow can language-conditioned robots preserve executable actions under irrelevant wording while composing multiple changed constraints correctly?A robot should ignore wording changes that leave the task unchanged but revise its action when task requirements change. Combining several requirements into one executable decision makes this distinction difficult.
AI
Evaluation & Benchmarks
Multimodal Models
Reasoning
Robotics
Latest papersRecent research connected to this question, newest first.One Word, Different Action: A Real-Robot Benchmark for Language-Conditioned Embodied ReasoningThe evidence concerns a real-robot benchmark using physical decision states, executable actions, task-preserving and task-changing instruction pairs, multi-constraint reasoning, and real-RGB grounding. Reported experiments find modern models near saturation for isolated instruction changes but noticeably less reliable when multiple constraints must be integrated.research paper · Sep 4, 2026
Related questions
How can real-time social robots reason over long-term context and decide when to act without disrupting fluent multimodal interaction?How can vision-language-action systems reliably execute long-horizon manipulation while tracking state and conditional dependencies?How can vision-language-action policies follow execution details beyond a robot task’s goal?How can LLM orchestrators preserve continuous state when collaborating with non-language agents?