Residual Rotation Correction using Tactile Equivariance

However, the high cost of tactile data collection makes sample efficiency the key requirement for developing visuotactile policies. We present EquiT ac, a framework that exploits the inherent SO(2) symmetry of in-hand object rotation to improve sample efficiency and generalization for visuotactile policy learning. EquiT ac first reconstructs surface normals from raw RGB inputs of vision-based tactile sensors, so rotations of the normal vector field correspond to in-hand object rotations. An SO(2)- equivariant network then predicts a residual rotation action that augments a base visuomotor policy at test time, enabling real-time rotation correction without additional reorientation demonstrations. On a real robot, EquiT ac accurately achieves robust zero-shot generalization to unseen in-hand orientations with very few training samples, where baselines fail even with more training data. T o our knowledge, this is the first tactile learning method to explicitly encode tactile equivari-ance for policy learning, yielding a lightweight, symmetry-aware module that improves reliability in contact-rich tasks.