Robotic Grasp/Motion Planning for Efficient Packing Assuming Multiple Grippers

Abstract

It is challenging for a robot to densely pack various types of objects into a box because the packing posture of an object depends on the grasp configuration of the gripper. In this study, assuming multiple grippers, we propose a method for automatically planning the motion of a robot that packs objects into a box , considering whether the grippers can grasp the objects and pack them in a box. The proposed method first determines the appropriate gripper candidates for grasping each object. The initial solution is then generated by arranging the objects in random order and placing them in the available space in order, considering the gripper for each object. To increase the filling rate of each container and reduce the final number of required containers, the system repeatedly exchanges and replaces the objects in each container. Finally, the grasping and placement postures using the appropriate gripper are obtained for each object position, based on which the packing motion is planned. Through simulation and experiments, we determine the container and the position where the object is to be inserted, and confirm that the object can be placed in a suboptimal position by planning the grasping motion of the robot without collision using a gripper that can grasp the object.