Reproduce “Hidden Variables Behind Real-Robot Data Quality: Lessons from Hundreds of Thousands of Hours” 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
Maniformer Real-World Manipulation Team, Che Tsang Calvin, Xingyu Chen, Bowen Jiang, Naixiang Gao, Xiongfeng Cai, Enlin Gu, Ruiying Li and 34 more
The authors examine how robot teleoperation setups shape demonstrations and learned policy behavior. In an out-of-view block-pick task, having the robot search with its camera before choosing an arm raised arm-choice accuracy from 50% and 53% to 100%; detour-contaminated demonstrations also increased policy path length. These findings come from selected tasks, robot configurations, and policy tests, not general causal estimates.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.