Reproduce “In-context Robot Learning Made Simple: A Democratized Recipe for Manipulation Tasks” 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
Minxing Li, Minghao Han, Weizhi Zhao, Hanwen Wang, Xiangshuo Liu, Shuyao Shang, Jingxiang Zhou, Mingchao Sun and 5 more
SimpleICL trains robots to use human video demonstrations by distinguishing task intent—action, object, order, and manipulation—from changes in position, motion, and scene. It creates contrasting examples and pairs demonstrations with executions across related scenes. On eight unseen real-world tasks, its Hard-level average success rate was 0.68, versus 0.42 for Fast-WAM and 0.31 for π0.5.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.