Abstract

This paper develops a receding horizon optimisation scheme for integrated behaviour decision-making and trajectory tracking to ensure the dynamic collision avoidance of an autonomous surface vessel (ASV). We apply Q-learning to make behaviour decisions because manoeuvring habit requirements and COLREGS must be satisfied in dynamic collision avoidance scenarios. The heading course and vessel speed are considered to transmit the outcome of behaviour decision-making to a trajectory tracking model predictive control (MPC) controller. Next, a trajectory tracking nonlinear controller for the ASV is developed within the MPC framework, in which a set of nonlinear constraints is designed for collision avoidance. Specifically, depending on the heading course and vessel speed, the collision avoidance constraints can be switched to allow the controller to execute the behaviour decision. Simulation results verify the effectiveness of the proposed receding horizon optimisation scheme.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call