TAMPAgent: Agentic Belief-Space Planning for Mobile Manipulation
TAMPAgent: Agentic Belief-Space Planning for Mobile Manipulation
This research presents TAMPAgent, an agentic planning framework that unifies Large Vision-Language Model (LVLM) reasoning with Task and Motion Planning (TAMP) constraint satisfaction. The system operates in open-world environments with partial observability by formulating parameterized task plans and binding valid geometric constraints via a tool API. Evaluated on Rainbow Y1 and Unitree G1 platforms, TAMPAgent demonstrates superior performance over direct tool-calling baselines in complex household mobile manipulation tasks, highlighting the necessity of integrating structured planning for unstructured environments.
The paper introduces TAMPAgent, a framework combining Large Vision-Language Model reasoning with Task and Motion Planning to handle geometric uncertainty in partially observable mobile manipulation tasks.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
Source：arXiv Robotics — research abstracts · arxiv.org