Tactile Intelligence: The Next Frontier in Robotic Dexterity
Proception, a startup specializing in robotic hands, has secured $11 million in funding after settling a trade secret dispute with Tesla. The company focuses on the high-fidelity tactile data required for human-like manipulation.
Fine motor control remains the "holy grail" of robotics. While we have mastered bipedal locomotion and warehouse navigation, the ability for a robotic hand to pick up a strawberry without crushing it—or use a screwdriver in a cramped space—remains incredibly difficult. Proception, a startup that recently emerged from a legal battle with Tesla, is doubling down on a unique approach to this problem: high-fidelity tactile sensing coupled with imitation learning.
The company’s $11 million raise underscores the growing investor interest in "end-effector" technology. Most industrial robots today use simple grippers or vacuum suctions, which are useless for the complex assembly tasks required in modern electronics or home care. Proception’s technology aims to mimic the density of human nerve endings, providing the AI with the sensory feedback loop necessary to adjust grip strength in real-time.
This tactile data is then fed into neural networks trained on thousands of hours of human demonstrations. This specific branch of robotics is moving away from purely visual-based control, which can be fooled by shadows or occlusions, and toward haptic-based control. By focusing on the hand, Proception is enabling the next generation of humanoid robots to actually do work, rather than just move through space.
Source: TechCrunch
