A Breakthrough in Robotic Object Manipulation Using Touch Alone

Engineers at the University of California San Diego (UCSD) have developed a remarkable robotic hand that can rotate objects based solely on touch, eliminating the need for visual input. Inspired by how effortlessly humans handle objects without relying on vision, the team equipped a four-fingered robotic hand with 16 touch sensors spread across its palm and fingers. Each sensor detects whether or not an object is touching it, providing simple binary signals for the robotic hand to perform in-hand rotation.

This approach is different from other methods that employ a few high-resolution touch sensors placed primarily at the fingertips of the robotic hand. Xiaolong Wang, the lead researcher and professor of electrical and computer engineering at UC San Diego, explained that these previous methods have limitations. They often minimize contact with the object, limiting the hand’s sensing ability. Moreover, high-resolution touch sensors that provide texture information are difficult to simulate and expensive, making them impractical for real-world experiments.

The team trained their system using virtual simulations of a robotic hand rotating various objects, including those with irregular shapes. The system determines which sensors are in contact with the object during rotation, as well as the current and previous positions of the hand’s joints. With this information, the system guides the robotic hand to the appropriate joint movements in the next time point.

Tests with real-life objects that were unfamiliar to the system demonstrated the effectiveness of the robotic hand. It successfully rotated objects like tomatoes, peppers, cans of peanut butter, and even a toy rubber duck with challenging shapes. Although more complex objects took longer to rotate, the robotic hand displayed versatility by rotating objects around different axes.

The researchers are now focused on expanding their approach to more complex manipulation tasks, such as catching, throwing, and juggling. By enabling robotic hands to master in-hand manipulation, they hope to broaden the range of tasks robots can perform. This breakthrough in robotic manipulation not only represents a significant step forward in the field of robotics but also has the potential to empower robots to manipulate objects in the absence of light or in visually challenging environments.