Path-following and LiDAR-based obstacle avoidance via NMPC for an autonomous surface vehicle

This paper presents a path-following and collision avoidance system for autonomous surface vehicles based on nonlinear model predictive control. The proposed strategy is capable of following a desired path while maintaining a commanded velocity, whereas it can diverge from the references to safely perform maneuvers to elude unexpected obstacles. A nonlinear dynamic model of the vehicle is applied to predict the vehicle states into a finite horizon. Furthermore, a LiDAR sensor located at the front of the boat is employed for local object detection. Here, the high-performance optimal control framework, acados, is used to solve the optimization problem onboard. Real-time numerical simulations and field experiments demonstrate the effectiveness of the proposed approach against multiple buoys.