Reproduce “TAMPAgent: Agentic Belief-Space Planning for Mobile Manipulation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Utkarsh A. Mishra, Alperen Degirmenci, Yongxin Chen, Danfei Xu, Tucker Hermans, Caelan Garrett
TAMPAgent guides a robot through manipulation by combining an LVLM agent’s hypotheses and plans with a task planner and perception/manipulation tools. The planner translates actions into geometric constraints, and tools certify values for each action before execution. On simulated RBY1 kitchen tasks, absolute success averaged 77.2%, versus 57.2% for CaP and 61.1% for GaP; this does not establish general reliability.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.