Reproduce “HandsOn: A Large-Scale Simulation Benchmark for Generalizable Dexterous Manipulation” 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
Tianwei Luo, Yiming He, Hongjin Wu, Ting Pan, Debin Ren, Haorui Guo, Junlin Yang, Yuxin Zuo and 9 more
HandsOn introduces a benchmark and data platform for evaluating dexterous manipulation beyond grasping. It provides 18,000 simulated human-demonstration trajectories and tests learned policies on 50 demonstrated tasks and 16 evaluation-only transfer tasks. The best trained checkpoint, OpenWAM, achieves 29.1% task-macro success on Core-50 but 0.4% on Transfer-16, showing a sharp transfer boundary under this protocol.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.