Reproduce “BAT-WAM: Bridging Action and Touch in World-Action Models for Dexterous Hands” 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
Dong Fang, Feixuan Huang, Yuanxin Zhong, Yuxiang Gao, Qi Song, Yunlong Wang, Xiaosong Jia, Yu-Gang Jiang
BAT-WAM reconstructs missing tactile information by separately encoding touch and hand-joint state, then combining them with RGB context through a shared bridge. In local-mask completion, its Action+Image input achieves H-RMSE 0.0137, compared with 0.0224 for the image-conditioned HandTouch comparator. In seven real-robot tasks, it averages 79.3% success, but does not lead on every task.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.