An Optimization-based Motion Planner for a Mobile Manipulator to Perform Tasks During the Motion

Jingren Xu, Yukiyasu Domae, Weiwei Wan, Kensuke Harada

Abstract: In this paper, we present an optimization-based motion planner to plan a locally time-optimal whole-body motion of a nonholonomic mobile manipulator, to pick up objects while simultaneously moving the manipulator and the base. The simultaneous motion further reduces the operation time of the picking tasks. What distinguishes our planner from the common motion planners, which plan the motion between two configurations, is that our planner considers performing tasks, such as grasping an object, during the motion. We formulate the time-optimal motion planning as an optimization problem. One of the major difficulties is finding an appropriate representation of the constraints for the tasks during the motion, since the time and configuration of the robot at the moment of performing the task are unknown. To address this issue, we propose a novel formulation of the optimization variables such that constraints arising from the tasks are smooth and differentiable, which is essential for obtaining a feasible solution using an NLP solver. We present the preliminary numerical result of the proposed planner, which shows that our planner can obtain a feasible trajectory that satisfies all the constraints.