Model Predictive Control of Nonholonomic Mobile Robots with Backward Motion

Currently, most of model predictive control (MPC) of moible robots are designed based on forward movements, which is not efficient when the robot in certain initial positions. In this paper, a novel model predictive controller is designed to solve the regulation problem of a nonholonomic wheeled mobile robot with backward motion when its initial position in the first and second quadrants of Cartesian coordinates. System state variables selection and corresponding kinematic models in polar coordinate are defined to characterize backward movements. The terminal state cost function and terminal region together with the local controller are designed to guarantee the stability of the optimization problem (OP). Comparison studies show that the proposed MPC algorithm outperforms conventional ones.