A Robotic Hand Surpassing Human Capabilities in Dexterity and Functionality

The human hand is often viewed as the pinnacle of dexterity. Yet, its asymmetric shape and unique thumb largely limit its dexterity. Artificial and robotic hands have often departed from anthropomorphic design. With two or three fingers distributed in a uniform manner, traditional industrial robotic hands aimed at preserving symmetry, and facilitating manipulation. They remain, however, extremely far from human dexterity, capable solely of picking and placing tasks, one object at a time. We present a reversible robotic hand that tackles this challenge by uniting multi-object grasping and crawling locomotion in a single device. The hand employs an identical-finger, symmetric design optimized through a multi-layer framework that combines optimization (for exploring diverse grasp configurations) with constraint-based methods. As a result, the hand can detach from the arm, crawl to retrieve multiple objects beyond normal reach, and reattach while securely holding them. This integrated approach expands conventional manipulative capabilities, enabling tasks that surpass human and traditional robotic hands in certain scenarios, such as grasping multiple items, while simultaneously walking, using a single hand in place of two for manipulating tools. By bridging the gap between stationary manipulation and autonomous mobility, our design opens new possibilities for industrial, service, and exploratory robotic applications.