Nonlinear Model Predictive Path Following for an Unmanned Surface Vehicle

This paper presents a two-level nonlinear model prediction control (NMPC) approach for autonomous path following control of under-actuated unmanned surface vehicles with constraints. It performs proper path replanning by taking into consideration of the dynamic constraints and input constraints of the USV in the high-level NMPC, and low-level NMPC solves the control inputs for the USV to follow the new path. The results show that the proposed two-level NMPC approach is able to reduce the change rate of control and avoid excessive control inputs compared with one-level NMPC approach, and the average computation time for each iteration is reduced, which indicates that two-level NMPC approach is more promising for real-time implementation.