Reproduce “Ego Guidance: Enhancing Robot Policy Learning with Action-Free Human Egocentric Videos” 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
Lin Long, Tianren Ma, Zhong Chu, Di Wu, Maosong Cao, Qixiang Ye, Junbo Zhao, Shanghang Zhang
Ego Guidance trains a vision-language backbone with robot-action prediction and future-image-feature prediction, letting human videos without action labels shape its representation. On LIBERO-Plus, SigLIP2 features achieve 76.2% average success across seven shift dimensions, versus 60.0% for Wan-VAE latents. OOD success rises with more Open-AoE video, while in-domain success changes little; raw video underperforms robot-only, showing benefits depend on curation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.