Reproduce “ReRead: Wrong Predictions, Useful Representations - Correcting Frozen Robot Policies by Re-reading Their Own Representations” 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

Xu Su, Shunlong Xiao, Qintao Yi, Mingkuan Zhao, Xinmiao Yu, Yizhe Li, Zihan Wang, Bokui Chen

ReRead corrects frozen robot policies by reading their stored representations after they propose an action. Its corrector uses current and two previous caches, robot state, and temporal information to predict a residual added to the proposal; it is trained on offline demonstrations. On RoboTwin 2.0 Hard11, Fast-WAM’s average success rises from 73.45% to 81.45%, in this selected benchmark subset.

Summary by alphaXiv.

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