Reproduce “HITCH: Teaching a Vision-Language Model to Operate a Humanoid through a Whole-Body Harness” 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
Qingzhou Lu, Chenxin Liu, Xinming Wang, Liu Cao, Yanjiang Guo, Mengdi Xu, Jianyu Chen
HITCH lets a vision-language model control humanoid tasks by measuring image targets in 3D and issuing whole-body commands for a planner and learned tracker. The authors train it on teacher demonstrations across five task categories, then test transfer without task-specific trajectories. It reaches 51.1% on SIMPLE and 42.5% on four HumanoidArena tasks, and succeeds in two tasks on a real G1.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.