Everything is under control: Energy-saving robot arm learns how not to drop the ball

Grasping objects of different sizes, shapes and textures is easy for humans, but it is challenging for robots. A team from the University of Cambridge in the UK has designed a low-cost, energy-efficient, flexible 3D-printed robot hand that operates with wrist movements and "skin" sensations to not only grasp a range of objects but also prevent them from falling. The research results were published in the recent journal Advanced Intelligent Systems.

The researchers have designed a low-cost, energy-efficient robotic hand that can grasp a range of objects without dropping them, using only wrist movements and "skin" sensation.
Image credit: University of Cambridge

The hand is trained to pick up different objects and can predict whether they will fall using information from sensors placed on its "skin." This passive motion makes the robot easier to control and more energy-efficient than robots with all-electric fingers.

Image credit: University of Cambridge

The researchers say their adaptable design could be used to develop low-cost robots that are capable of more natural movements and can learn to grasp a variety of objects.

The human hand is incredibly complex, and reproducing all of its dexterity and adaptability in a robot is a huge research challenge. Most of today’s advanced robots are incapable of performing manipulation tasks that a small child can do easily. For example, humans instinctively know how much force to use when picking up an egg, but this is a challenge for robots: too much force and the egg might break; too little and the egg will fall from the hand. In addition, a fully actuated robotic hand, with motors for every joint in every finger, requires a lot of energy.

Researchers at the Bio-Inspired Robotics Laboratory in the Department of Engineering at the University of Cambridge are trying to develop potential solutions to both problems: robotic hands that can grasp a variety of objects with just the right amount of force, using the least amount of energy.

The researchers used a 3D printed anthropomorphic hand implanted with tactile sensors so that the hand could feel what it was touching. The hand was only capable of passive, wrist-based movements. The team conducted more than 1,200 tests on the robotic hand, observing its ability to grasp small objects without dropping them. The robot was initially trained using small 3D printed plastic balls and grasped them using predefined movements learned through human demonstrations.

The robot learned what kinds of grasps would be successful through trial and error. After completing training with a ball, it tried grasping different objects, including a peach, a computer mouse, and a roll of bubble wrap. In these tests, the robot hand successfully grasped 11 of the 14 objects.

In the future, the system could be expanded in a number of ways, such as by adding computer vision capabilities, or teaching the robot to exploit its environment, which would enable it to grasp a wider range of objects.