Robots need precise instructions to grab things. Researchers working in the Computational Robot Design & Fabrication Lab at EPFL’s School of Engineering explain this challenge. Humans use soft interactions between their hands and objects. These interactions help them grab things easily. The researchers have explored how robots can use softness to grab objects. In robotics, softness refers to materials that bend or squish. This softness is called compliance. The researchers look for ways to use compliance to make robots better at picking up things.
The researchers created a robotic hand called ADAPT. ADAPT stands for Adaptive Dexterous Anthropomorphic Programmable sTiffness. This hand uses simple materials like silicone strips. Silicone wraps around a mechanical wrist and fingers. The hand also has spring-loaded joints. A bendable arm supports the hand. These features make the hand compliant. The hand automatically adjusts its grip to hold different objects. It doesn’t need detailed programming for each grasp. Researchers call this self-organized grasping. In tests, the ADAPT hand picked up 24 objects with a 93% success rate. Its grasps were 68% similar to human grasps. The study appears in Nature Communications Engineering.
Softness simplifies robotic control
A typical robotic hand needs many motors for each joint. The ADAPT hand has 20 joints but only 12 motors. These motors sit in the wrist. Springs and silicone skin handle the rest of the control. Springs can adjust to be stiffer or looser. Silicone skin can be added or removed. This setup tunes the hand’s softness. The hand follows just four basic positions to lift objects and adapts on its own without extra instructions. This method is called open-loop control. The hand grasped objects like a bolt or a banana using the same motion.
The researchers want to use the mechanical intelligence of body parts like skin and joints. This differs from brain-based intelligence. Developing robots that mimic human actions is hard. Compliance helps robots handle uncertain situations. The lab now adds sensors and artificial intelligence (AI) to the hand. Sensors in the silicone skin provide feedback. This mix of softness and control could make robots work better in unpredictable or human-designed spaces.
