Trajectory tracking of autonomous vessels using model predictive control ⋆

Autonomous surface vehicles are with increasing popularity being seen in various applications where automatic control plays an important role. In this paper the problem of twodimensional trajectory tracking for autonomous marine surface vehicles is addressed using Model Predictive Control (MPC). At each time step, the reference trajectories of a vessel are assumed to be known over a finite time horizon; the MPC controller computes the optimal forces and moment the vessel needs in order to track the trajectory in an optimal way. Based on a horizontal 3 degrees of freedom nonlinear scaled vessel model, we present both nonlinear MPC (NMPC), which solves a constrained multi-variable nonlinear programming problem, and linearized MPC (LMPC), which solves a constrained quadratic programming problem through on-line iterative optimization. In the latter case, the model used in LMPC for prediction is obtained from a successive linearization of the nonlinear vessel model. Comparisons on performance and computational complexity of the two approaches are presented. The effectiveness of the MPC formulations in vessel trajectory tracking, especially the ability of explicitly handling constraints, is demonstrated via simulations.