Tactile Adaptive Robot Paper Wins Best Paper

Scobleizer · x · 2026-07-18

OmniTacTune won the Best Paper Award at the Tactile Sensing for Robotic Foundation Models Workshop at RSS 2026.

This work focuses on a specific issue: while visual policies can learn task structures and motion priors from teleoperated robot demonstrations and human videos, the lack of paired tactile data often leads to failures in contact-intensive, last-few-millimeter manipulations. The authors propose using tactile residuals to adapt pre-trained visual policies, enabling the robot to acquire contact-phase capabilities without relying on inherently available tactile data.

Original post →

More from Embodied

Embodied channel →