Embedded Model Predictive Control for Enhancing Tracking Performance of a Ball-and-Plate System

Implementing model predictive control (MPC) on a limited-performance microcontroller has always been a great challenge. In this paper, we propose an MPC implementation method that can run on a microcontroller by improving the calculation efficiency and address applied to the ball-and-plate system to enhance tracking performance. First, we convert the MPC quadratic programming problem, which causes difficulties in implementing MPC, into an equivalent nonnegative least-squares problem so that we can build the solver in a C program easily. Subsequently, we separate the offline calculation from the online process, which drastically reduces the calculation time. Finally, we implement the proposed MPC in a C program to run it on a Nucleo-32 microcontroller, apply it to a tracking control problem using a laboratory-built ball-and-plate system, and then explain the improved tracking performance compared with the conventional control methods.