Reproduce “Learning Heterogeneous Tactile Representation for General 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
Bing Han, Yichuan Peng, Haoyu Dong, Yuanmeng Hu, Yongyi Jia, Fangjing Wang, Kecheng Zheng, Yiming Zhang and 11 more
The authors introduce LingBot-Touch, which maps different tactile sensors’ readings into a shared representation of contact that robot policies can reuse. Its frozen encoder supplies features alongside policy inputs. Across 26 real-robot tasks, adding these features to LingBot-VLA raises mean success from 48.3% to 73.1%, versus 56.5% with raw touch; gains are not universal, and evaluation covers two hands and fingertip-focused data.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.