Manipulability-Guided MPC with Repulsive Potential Fields for Mobile Manipulator Whole-Body Control

To address the insufficient flexibility and coordination in traditional hierarchical control strategies for wheeled mobile manipulators, this paper proposes a whole-body real-time control approach under a model predictive control (MPC) framework. Aiming for synergistic motion between the mobile platform and manipulator, a high-degree-of-freedom integrated kinematic model is established by equivalently representing the mobile platform as a virtual joint. Within the MPC framework, a multiobjective cost function is designed to simultaneously minimize pose tracking errors, enforce obstacle avoidance via repulsive potential fields, and integrate a manipulability optimization metric with dynamic weight adaptation mechanisms to refine motion planning. The kinematic model is implemented and computationally solved as a nonlinear programming problem using the CasADi optimization toolkit, while a Gazebo-based simulation environment is developed for validation. The simulation results demonstrate that the proposed method accomplishes the primary tasks of target tracking and collision avoidance while optimizing the motion process through manipulability optimization and dynamic weighting.