Toru Lin
Co-author focused on dexterous manipulation sim-to-real research at UC Berkeley.
“This paper solves a critical problem in tactile sim-to-real transfer — how to extract rich contact information from touch sensors without the signal collapsing when you move from simulation to hardware. The answer is a physics-grounded intermediate representation called Center-of-Pressure (CoP) that achieves zero-shot sim-to-real transfer on a 16-DOF hand doing precision manipulation, with no real-world training data required.”
Source→“On peg-in-hole insertion across six geometries, CoP achieved 78% overall success rate vs. 53% for binary contact and 48% for raw taxels (Table 1).”
Source→“Co-authored several prior papers on sim-to-real dexterous manipulation including visuotactile skills with multifingered hands [30] and sim-to-real for humanoid dexterous manipulation [28].”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.