Reproduce “HuGe: Scaling Humanoid Generalist Vision-Language-Action Models with Egocentric Experience” 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
Yang Liu, Jiacheng Liu, Hengtao Li, Yuheng Huang, Zijian Yang, Ji Dai, Maoxin Wang, Yi Gao and 9 more
HuGe adapts existing vision-language-action policies for humanoid manipulation by mapping human and robot motion into tokens decoded by a pretrained whole-body controller, then training through egocentric pretraining, humanoid adaptation, and task-specific fine-tuning. In HumanoidArena, HuGe-GR00T reaches a 66.67% unweighted average across seven tasks, versus 53.81% for a baseline score reported in prior work, not a new head-to-head run.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.