Robot that Can Pick and Sort comes with Recognition Algorithm 


Photo Credit via Pixabay


Many people dislike unpacking their groceries because of the tedious effort needed to feel around the bags and taking them out. Researchers at Massachusetts Institute of Technology and Princeton University developed a robotic system that may help people in picking and sorting items. 

The robotic system labeled "pick-and-place" is comprised by a standard industrial robotic arm equipped with a custom gripper and a suction cup. The robot can grasp objects using an algorithm to determine the best way on how to pick up random clutter. For the sorting, the robot captures images of a picked item in multiple angles using a set of cameras and an image-matching algorithm, then the robot compares it to a library of other images that closely match the picked item. If it matches, the robot separates that item and places it in a different bin. In simple terms, the robot follows the "grasp first and then recognize" flow and the researchers find it more effective than pick-and-place technologies. 

“This can be applied to warehouse sorting, but also may be used to pick things from your kitchen cabinet or clear debris after an accident. There are many situations where picking technologies could have an impact,” said Alberto Rodriguez, a professor of mechanical engineering at MIT. 

Rodriguez has been working on robots that feature adaptability, flexibility, and intelligence to overcome the flaws of most picking robots used in industries. Standard picking robots are usually designed with one repetitive task like pick and place. In the study, the researchers trained the new robotic arm to pick novel items from a cluttered bin using only one grasping capability. It can either suction on the object sideward or vertically, or grip it vertically and then use a flexible spatula to slide between the target object and the wall. 

After several trials and errors, they found that the robot had a 54 percent success rate using the suction while it had a 75 percent success rate using the gripper. The system was able to recognize novel objects with 100 percent accuracy. 

The researchers have already installed tactile sensors on the robot to train it to pick and sort using a new source of information.