Neural Dynamics-Based Model Predictive Control for Mobile Redundant Manipulators With Improved Obstacle Avoidance
Jingkun Yan, Mei Liu
Abstract
The trajectory-tracking problem of mobile manipulators is worth investigating because of its widespread appearance in industry and academia. This article proposes a model predictive control (MPC) scheme aimed at controlling a mobile manipulator and enabling it to track a given trajectory in an environment with obstacles. The proposed MPC scheme is simultaneously constrained by the joint angle limit and joint velocity limit of mobile manipulators. These constraints take effect in their original form, thereby preserving the original feasible region of joint angles, which differs from the previous studies that may diminish the feasible region of joint angles. Additionally, an improved obstacle avoidance (IOA) method is proposed and integrated into the MPC scheme to ensure the collision-free operation of the mobile manipulator in an environment with obstacles. Compared to the previous studies, the proposed IOA method expands the direction range of the escape velocity and enables the mobile manipulator to achieve satisfactory tracking accuracy. Last, a neural dynamics solver is developed to realize online solving of the MPC scheme. A comprehensive set of computer simulations and physical experiments are conducted, involving the mobile manipulator carrying out trajectory-tracking tasks in various environments with different obstacles. The results from both simulation and experimentation corroborate the superiority of the proposed neural dynamic-based MPC algorithm with the IOA method.