Two new developments in the field of grippers can improve the accuracy and capability with which robotic arms grip objects. They involve a special type of gripper inspired by the carnivorous plant and a system that helps robotic arms determine how to pick up and manipulate an anomalous object.
Both developments come from MIT. Called the 'origami robot gripper' by the university, the gripper consists of a conical origami structure that is placed over an object and then 'folds in', causing the material to fold around the object. This offers all kinds of advantages. For instance, the force the gripper exerts on an object is distributed over a wide area, allowing the gripper to pick up even delicate objects without causing damage.
Flexibility
Gripper also gives robot arms a lot of flexibility. For example, the gripper can pick up a variety of objects with all kinds of shapes, as the shape of the gripper automatically adjusts to the shape of the object after closing. For example, the gripper can pick up soup cans, hammers, wine glasses, drones and individual broccoli florets. The gripper can pick up objects up to 70% of its own diameter. It could also lift bottles weighing almost 2 kilograms.
"Companies like Amazon and JD want to be able to pick up a wider range of delicate or irregularly shaped objects but cannot do so with finger-based grippers or grippers with suction cups," said Shuaguang Li, a postdoctoral fellow at MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) and Harvard's John A. Paulson School of Engineering and Applied Sciences. "Suction cups cannot pick up things with holes - and they need something much stronger than grippers with soft fingers." With the origami robot gripper, the researchers have developed a solution to this.
More information on the origami robot gripper can be found here.
Keypoint Affordance Manipulation
The second development also comes from MIT'S CSAIL. It involves a system called 'Keypoint Affordance Manipulation (KPAM), which enables robots to accurately pick up anomalous objects. The system does this by scanning an object and imaging coordinates - also called keypoints - from it.
KPAM can increase the deployability of robotic arms. For example, robotic arms typically cannot handle objects they have not seen before very well. This results in limitations, partly because robotic arms cannot properly manipulate products that show abnormalities. In practice, this means they are especially suited to picking up objects in highly structured environments in an automated way. KPAM offers a solution and gives robot arms much more flexibility to deal with deviations.
Enabling useful manipulations
The coordinates collected by KPAM give robots a better understanding of the object they want to pick up. The researchers involved claim that the system only needs three coordinates to pick up relatively uniform objects such as a coffee cup. Objects such as slippers and shoes require six coordinates. "By understanding just a little bit more about the object - the location of some key points - is enough to enable a wide range of useful manipulations," explains Russ Tedrake, a professor at MIT. "Robots can pick up almost anything, but if they haven't seen the object before they usually can't put it down in a meaningful way."
In a video, the researchers demonstrate how KPAM works. This includes how the robot is able to hang coffee cups a hooks and place shoes in a shoe rack.
More information on KPAM is available here.
Author: Wouter Hoeffnagel
Source: MIT CSAIL
