Researchers at the University of Cambridge have developed a cost-effective and energy-efficient robotic hand that can grasp various objects without dropping them. Unlike traditional robots with motorized fingers, this soft, 3D printed hand relies on passive movement controlled by the wrist and tactile sensors on its “skin.” By analyzing the sensor data, the robot can predict if it will drop an object and adjust its grip accordingly.
This innovative design could revolutionize the field of robotics by enabling the development of more affordable and versatile robots capable of natural movements and learning to grasp different objects. The researchers published their findings in the journal Advanced Intelligent Systems.
The human hand is incredibly complex, and replicating its dexterity in a robot poses significant challenges. Most advanced robots today struggle with manipulation tasks that even small children can perform effortlessly. Picking up fragile objects, like an egg, requires a delicate balance of force. Too much force can lead to breakage, while too little can result in dropping. Additionally, fully motorized robot hands consume a considerable amount of energy.
At the Bio-Inspired Robotics Laboratory in Cambridge, Professor Fumiya Iida and his team have been exploring solutions to these challenges. They aimed to create a robot hand capable of grasping objects with the right amount of force while conserving energy.
Their approach involved utilizing the passive movement of the wrist and implanting tactile sensors on a 3D-printed hand. The sensors allowed the robot to perceive what it was touching. Despite only having wrist-based movement, the hand demonstrated impressive performance in over 1200 tests, successfully grasping small objects without dropping them.
The robot learned through trial and error, adapting its grip based on the sensor data. The team initially trained it using 3D printed plastic balls, then tested its ability to grasp a variety of objects, including a peach, a computer mouse, and a roll of bubble wrap. Out of 14 objects, the hand successfully grasped 11.
This passive design offers several advantages over fully actuated robot hands. With fewer parts and sensors, it is easier to control, provides a wide range of motion, and simplifies the learning process. The researchers believe that by adding computer vision capabilities and teaching the robot to utilize its environment, the system could be further enhanced to grasp an even wider range of objects.
This research was supported by UK Research and Innovation (UKRI) and Arm Ltd. Professor Fumiya Iida is affiliated with Corpus Christi College, Cambridge. The development of this robotic hand opens up exciting possibilities for the future of robotics, where low-cost and energy-efficient robots can perform complex tasks with ease.