Reproduce “Perceptive Humanoid Motion from Simulation” 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
Siyuan Hu, Abulikemu Abuduweili, Yue Meng, Shipeng Liu, Quentin Leboutet, Chen Chen, Matthias Müller, Vladlen Koltun
HALO trains a humanoid robot to turn language requests and camera views into whole-body actions. It uses simulated homes to generate paired vision-language-motion examples while people still author task prompts and motion keyframes. In six unseen, unmapped real-world homes, it achieves 91% mean success across six tested single-task skills, not longer sequences. Its stack combines planning, generation, and tracking.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.