Model Predictive Control: Systematic Design Tailored to Nonholonomic Systems
Nonholonomic systems are ubiquitous in applications such as transportation and robotics, yet their control remains challenging. In order to achieve asymptotic stability of a desired equilibrium, model predictive control (MPC) schemes without stabilizing terminal conditions have recently been proposed for nonholonomic kinematic systems. However, this approach still required lengthy system-specific calculations. This paper eliminates the need for these auxiliary calculations, as we propose a streamlined systematic design for MPC stage costs tailored to nonholonomic systems, utilizing the concepts of privileged coordinates and nonholonomic weights. Furthermore, the proposed design methodology is extended beyond purely kinematic models to a wider class of systems with non-negligible dynamics, and we highlight the geometric meaning of the resulting MPC stage costs. The systematic procedure is demonstrated using an articulated vehicle as an example.