Nonlinear MPC based Whole-body Control for Mobile Manipulation

In the control of mobile manipulators, maintaining the manipulability of the manipulator, the fast mobility of the base, and the coordination of all joints enduring challenges. This paper proposes a whole-body motion controller for a nonholonomic mobile manipulator based on Nonlinear Model Predictive Control. By imposing constraint conditions, the controller avoids joint position and velocity limitations during the process of guiding the mobile manipulator to the target points. The controller maximizes the ability of manipulator to move in any direction, optimizes the mobility of base, and simultaneously minimizes end-effector velocity tracking error through the objective function. In a simulated environment, comparative experiments are conducted between the proposed controller and a whole-body motion controller based on quadratic programming. The results demonstrate the superior performance of our controller, producing motions that enhance manipulability, minimize tracking error, and meet the requirements for rapid mobility when compared to the latter.