Reproduce “PyRUA-DataFarm: LLM Agent as a Cost-Efficient Teleoperator for Real-Ready Robot Data” 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
Kewei Zhang, Juncheng Ma, Guodong Lv, Duomin Wang, Feixiang Ruan, Ruiyang Si, Chuanrui Zhang, Yuxuan Chen and 6 more
DataFarm uses a coding agent to build a simulation task from a request and a calibration recording, then generates action-labelled episodes whose actions are rendered as synchronized camera views. The recording’s actions are not reused. On one dual-arm robot, farm-only policies achieved 87 successes versus 46 for teleoperation-trained policies in 100 trials across five tasks under unseen appearance conditions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.