Milp-based dual-arm motion planning considering shared transfer path for pick-up and place

Abstract

One of the most basic tasks that a dual-arm robot does is pick-up and place work. Pick-up and place work consists of tasks in which the robot carries objects from a start position (initial position) to a goal position. The following three important points should also be considered when the dual-arm robot does this work efficiently: (1) collision avoidance of the arms, (2) which arm should move an object, and (3) the order in which the objects should be picked up and placed. In addition, dual-arm robot has operation range constraints. Depending on the position relationship between a start position and goal position, unless both arms are used, the object may not be transferred to a goal position. In this paper, we define the transfer path which must use both arms as “shared transfer path”. First, we use mixed integer linear programming (MILP) based planning for the pick-up and place work to determine which arm should move an object and in which order these objects should be moved while considering the dual-arm robot’s operation range. Second, we plan the path using the rapidly exploring random tree (RRT) so that the arms do not collide, enabling the robot to perform efficient pick-up and place work based on the MILP planning solution. The effectiveness of proposed method is confirmed by simulations and experiments using the HIRO dual-arm robot.