Reproduce “Agentic RSR: Real-to-Sim-to-Real through Scene Reconstruction and Execution-Grounded Robot Policies” 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
Yihan Li, Yating Feng, Shengjiu Sun, Jianing Chen, Hao Ren, Bowen Yang, Weisheng Xu, Qiwei Wu and 2 more
Agentic RSR connects reconstruction, policy development, and robot execution in one task-level loop. It builds a simulation from a workspace, lets a policy develop through observations and retries, then transfers its experience to hardware, where the policy must reobserve and adapt. Across 18 tasks, 10 succeed in simulation and 8 also on hardware: paired transfer is 80.0%, while real success is 8/18.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.