Reproduce “Ego2X: Contact-aware Transfer from Egocentric Human Data to Any Robot” 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

Xinke Lyu, Bowen Yang, Zhicheng Wang

Ego2X transfers human video interactions to robots by recording object-relative contacts, interaction states, and force priors, then using them to refine geometry and guide robot motions. Across 25,549 contact slots in 79 EgoDex takes, it realizes 62.9% of target contacts, versus 29.6% for wrist-following control; verified data reaches 67.0% success on one simulated task, not real-world deployment.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.