Reproduce “MobileEgo Anywhere: Open Infrastructure for long horizon egocentric data on commodity hardware” 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

Senthil Palanisamy, Abhishek Anand, Satpal Singh Rathore, Pratyush Patnaik, Shubhanshu Khatana, Ekaksh Janweja

MobileEgo Anywhere provides a phone-based pipeline for collecting long egocentric activity with spatially tracked hands and hierarchical language labels. A head-mounted iPhone Pro records synchronized streams; STERA processes them into hand trajectories and labels. The dataset spans 200 hours across 584 sessions, with a 108-minute maximum. Mid-training reduces held-out open-loop hand-action prediction loss by 19%, not robot task success.

Summary by alphaXiv.

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