Cambridge University researchers have designed a robotic hand that learns not to drop the ball or hold a chopstick based on information provided by sensors in its human-like synthetic skin.
The robotic hand designed by Cambridge engineers can grasp a range of objects and not drop them by using just the movement of its wrist and sensors in its “skin.”
Researchers say that most of today’s advanced robots cannot manipulate tasks that small children can easily perform, such as picking up an egg without dropping or shattering it.
A human instinctively knows how much force to use when grasping objects of different sizes, shapes, and textures – and does it without using unnecessary energy.
Cambridge researchers have tried to recreate some of that dexterity by 3D-printing a soft robotic hand that cannot independently move its fingers but still carries out a range of complex movements.
In a video demonstration, the robot hand can be seen struggling to pick and hold objects like a ball but then “learning” to use the exact amount of force needed to grasp it without dropping.
The team behind the study, published in the journal Advanced Intelligent Systems, carried out more than 1,200 tests with the hand, using a predefined action obtained through human demonstration. The researchers 3D-printed the hand and implanted it with tactile sensors.
“The tactile sensors give the robot a sense of how well the grip is going, so it knows when it’s starting to slip. This helps it to predict when things will fail,” Dr. Kieran Gilday, first author of the study, said in a Cambridge University announcement of the research.
The hand picks objects without any actuation of the fingers. This “passive” movement saves significant energy that a fully actuated robot hand, with motors for each joint in each finger, would otherwise use.
“In earlier experiments, our lab has shown that it’s possible to get a significant range of motion in a robot hand just by moving the wrist,” study co-author Dr. Thomas George-Thuruthel was quoted as saying.
The adaptable design of the robotic hand could be used to develop low-cost robots capable of more natural movements, which could learn to grasp a wide range of objects, Cambridge University said.
More from Cybernews:
Subscribe to our newsletter