Reproduce “InfCoiL: Coordinated Planner-Controller Learning for Closed-Loop Physics-Based Human-Object Interaction” 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
Yude Zou, Yifei Yao, Yiwei Hao, Hanqing Wang, Xing Gao
InfCoiL closes the loop between motion planning and simulated humanoid control: it feeds executed body, object, and contact states into future plans. The authors alternately train planner and controller from rollouts, using a shared morphology interface across body shapes. On held-out tests, coordinated training achieves 81.19% success at 0.3-meter tolerance versus 71.99% for controller-only training with matched controller iterations, but not matched total computation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.