Data-Driven MPC for Trajectory Tracking Control of Wheeled Mobile Robots

The need for an explicit system model limits the applications of model predictive control (MPC). In this paper, two data-driven MPC algorithms are developed for the trajectory tracking control of wheeled mobile robots (WMRs). The algorithms use an implicit system model based on the Hankel matrices that are constructed from input and output data. The controller design is discussed for a linear equivalent model and the nonlinear model. To apply the data-driven MPC to the nonlinear model, a dynamic linearization strategy is proposed with a robust problem formulation to consider the modeling error explicitly. Simulation results show that the proposed data-driven MPC method could achieve satisfactory tracking accuracy while avoiding the explicit modeling of the system dynamics.