Reproduce “EmbodiedEvo: Self-Evolving Coding Agent For Physical Intelligence” 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
Wenyi Li, Yizhe Chi, Youjie Zheng, Chunyu Liu, Xiaoqiu Wang, Mingju Gao, Zheng Jiang, Deyao Hong and 3 more
EmbodiedEvo lets a robot coding agent preserve experience from one scene as executable Skills and diagnostic Lessons, then transfer them to a fresh agent in another scene. In matched RoboDojo-Sim development tests, Astra/xhigh success rose from 8.50% to 30.17%. Yet full evaluation Precision success was 13.25%, showing that inherited procedures do not ensure successful contact control.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.