Reproduce “X-WBC: A Cross-Embodiment Foundation Model for Humanoid Whole-Body Control” 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
Juntong Zhang, Chun Gu, Li Zhang
The authors introduce X-WBC, a humanoid whole-body controller trained jointly across robots. It aligns human motion, retargeted robot motion, and sparse virtual-reality observations, then combines a shared temporal Transformer with robot-specific sensing and action modules. On BONES-SEED training motions, G1 reaches 98.60% success versus 97.69% for a separately trained policy; this is in-distribution evidence, not unseen-robot control.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.