Motion planning for mobile manipulators

Abstract

A mobile manipulator, i.e., a manipulator mounted on a mobile platform, can perform various tasks over a large space that could not possibly be accomplished by a fixed-base manipulator by utilizing the mobility of the platform. Since the manipulator and the mobile platform has independent mobility there occurs a particular kinematic redundancy introduced by the mobility of the platform, which is basically different from that introduced by extra joints. This redundancy is quite desirable for dexterous motion in cluttered environments, however it complicates the motion planning problem significantly. Therefore, the study deals with the motion planning issues for mobile manipulators. We consider the point-to-point task, continuous- path task and task with unknown a-priori end-effector trajectory, which are essentially required in real application. Firstly, we propose motion planning methods for point-to-point task, which is defined as the configuration (position and orientation) of the end-effector with respect to the world frame. Considering the environment where the mobile manipulator works is generally cluttered and hostile, in order to test whether the desired task is executable or not without violating the imposed constraints, we propose a mobile manipulator executability test algorithm. We consider the general manipulator, the inverse kinematics of which is unknown. And we propose the method to obtain the optimal configuration for a single task if the desired task is executable. For the motion planning for multiple task which consists of a sequence of tasks, because the motion between tasks affects the performance of the multiple task, it should be considered. Therefore, we propose a motion planning method for multiple tasks with consideration of the motion feasibility. We formulate proposed three problems as an optimization problem. Because these problems are highly non-linear and the solution spaces are generally unconnected and nonconvex, poor solution i...