Tactile Sensing for Robotic Manipulation

This field studies how robots can use touch—such as pressure maps, shear, vibration, and contact geometry—to perceive interactions that cameras may miss. Researchers build optical, resistive, capacitive, and acoustic sensors, then use tactile feedback for tasks such as stable grasping, slip correction, delicate force control, and peg insertion. Learning approaches range from fast reactive controllers and tactile state estimators to vision-tactile policies, world models, and simulation-to-real training; benchmarks increasingly test different sensors, robot hands, and contact-rich tasks. A central challenge is making tactile representations and policies transfer across sensors and settings, since the value of touch depends on the task and simply adding tactile input does not guarantee better control.