Reproduce “ME-Dex 1.0: Bringing Heterogeneous Tactile Sensing into World Action Modeling” 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
Xuancheng Zhang, Xuetao Liu, Qianying Tang, Jizhe Wang, Runsheng Wang, Zhijing Cheng, Bochen Lin, Haoran Wen and 3 more
ME-Dex-1.0 makes touch a future prediction target alongside video and action, using three interacting experts, flow matching, shared attention, and representations for heterogeneous tactile sensors. In RoboTwin mixed training, it reaches 91.92% Random closed-loop success versus 90.22% for Fast-WAM+Tac. An ablation supports the design, but policies used one training seed.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.