Reproduce “Learning Quality-Controllable Robot Policies with Iterative Hindsight Relabeling” 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
Sinjae Kang, Dong Kyu Shin, Jaehong Kim, Kamin Lee, Hyoeun Kim, Kaixin Wang, Li Zhao, Seonghyeon Ye and 1 more
QCP lets users request execution quality by labeling successful trajectories with rule-based scores and conditioning a policy on those labels. It collects successful rollouts at the highest label, relabels data by measured quality, and retrains. On a conveyor, its highest-bin policy raises end-effector speed from 188.6 to 257.6 mm/s and throughput from 8.2 to 13.0 cans per minute versus BC.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.