The computer scientists at Rutgers University have developed artificial intelligence which enables them to control a robotic arm that can efficiently pack boxes, thereby reducing business time and money that gets invested in these tasks.
Kostas Bekris, the study’s senior author and an associate professor in the Department of Computer Science in the School of Arts and Sciences of the University in New Brunswick, said that with the help of this new venture they aim at achieving low-cost and automated solutions that can be deployed deftly. He further says that the key to this is in coming up with nominal yet effective hardware choices, and instead concentrating their focus on resilient algorithms and software.
Abdeslam Boularias and Jingjin Yu, both assistant professors in the department, formed a team with Bekris to handle the multifaceted issues regarding the robotic packaging system. They have thought of doing so with the help of select hardware, 3D vision and vigorous motion.
This peer-assessed study of the scientists was published at the IEEE (Institute of Electrical and Electronics Engineers) International Conference on Robotics and Automation (ICRA) where it was one of the finalists for the Best Paper Award in Automation. This study has much in common with the growing tendency of using robots to conduct logistics, retail and warehouse chores. There is a steady and speedy growth in the advancement of robotics because of the machine-learning algorithms which make way for continuous experimentation.
The packaging of unarranged boxes from a pile into tightly organized products has largely been a manual task for all these years. The Rutgers scientific team says that in automating such tasks, the people can focus on more important duties which will help prosper the companies’ competitiveness.
The Rutgers scientific team’s study mainly focused on taking objects from the bin and arranging them tightly in shipping boxes. After coming up with developed algorithms and software for the robotic arm, they created a suction cup that doubles itself and functions as fingers which push the objects in place. It also uses sensor data that pulls the objects towards a targeted area and pushes them together. Real-time monitoring is used in these operations to identify and overcome potential failures.