Reproduce “MoDex: Towards Universal Multi-Object Dexterous Grasping” 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
Jing Xiong, Jiale Xie, Haoqi Yuan, Chaoyi Xu, Zongqing Lu
MoDex adapts a shared reference grasping trajectory for two objects and selects a one-shot or staged strategy based on their spacing. In simulation, it achieves 61.24% success on unseen object geometries. A vision-based policy transfers to physical trials, succeeding 55% in proximal layouts and 45.5% in separated layouts. These results cover two-object tasks, not universal grasping.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.