6thSense

Robots today rely primarily on vision to identify and manipulate objects, but vision alone cannot tell a robot how much force or pressure to apply during a task like picking up a strawberry without crushing it. The company cites research claiming vision-only approaches succeed about half the time, while incorporating touch raises success rates above 80%. Teams working on robot learning and dexterous manipulation currently lack a way to capture force and pressure data directly, leaving a gap between what a robot can recognize and what it can safely grasp or handle.

6thSense makes two hardware products: Skin, a wearable tactile sensor that records every touch as force data, and Eye2 Pro, an egocentric camera synced to that touch data to capture first-person video alongside pressure readings. Together these let robots (or the models trained on this data) learn grip and pressure control from real tactile feedback rather than inferring force from vision alone. The company has posted tactile capture demos showing the hardware in use for tasks like picking up a box, a hammer, and a shoe.