Reproduce “TactileLab: Sim-to-Real Dexterous Robot Manipulation with Efficient Shear-Sensitive Tactile Simulation” 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
Yijiong Lin, Wenhao He, Toby Jack, Max Yang, Haoran Li, Nathan F. Lepora
TactileLab derives dense image-space contact-motion flow from rigid-body kinematics and combines it with contact-depth maps for manipulation policies, without simulating tactile-skin deformation. In real-world in-hand rotation with a LEAP Hand, depth-plus-flow averaged 1.95 ± 0.81 complete equivalent revolutions versus 1.70 ± 0.78 for depth alone; both averaged the 30-second trial limit. This result is limited to the tested protocol.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.