When Unimate, the world’s first industrial robot arm, showed off its skills by assembling a breakfast of buttered toast, coffee and champagne half a century ago, it seemed that robots were on the cusp of being able to carry out virtually any task we could perform with our hands. But things haven’t turned out to be so easy.
While Unimate’s every move was carefully coded, robots which have to deal with the uncertainties of the real world have proven less adept at being able to manipulate objects with the right amount of dexterity. A new soft robot gripper is here to help. Developed by researchers from the Massachusetts Institute of Technology (MIT) and Harvard University, the suction-based gripper resembles a kind of rubberized Venus flytrap.
“[The gripper] consists of two main parts: A flexible membrane as ‘skin,’ and an origami-inspired compressible structure as ‘skeleton,’” Daniela Rus, director of MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL), told Digital Trends. “The skin encloses the skeleton, forming a hemispherical flower-like gripper. When air is sucked out of the sealed skin, the skeleton is compressed to tightly hold around the external objects.”
By closing its cone-shaped end-piece over whichever objects it wants to move, the robot gripper can pick up a range of objects up to 100 times its weight. These objects can range in size and shape from drones and soup cans to something as delicate as a wine glass or single broccoli floret. Compared with rigid grippers, it has the advantage of being more flexible and robust. Compared with soft grippers, it is able to produce a stronger holding and grasping force. In other words, it represents the best of both worlds.
As Rus explained, the gripper could be useful for a range of possible applications, such as warehouse order picking or grocery packing.
“In the future, we plan to try different materials for the skin and skeleton, in order to optimize the strength, rigidity, and durability of the gripper,” she said. The team behind the gripper project also wants to incorporate computer vision elements so that it is able to see the world around it and learn to grasp specific parts of objects where necessary.