Intelligent planning with performance assessment for Autonomous Surface Vehicles

A mathematical model of agent-based decision-making of Autonomous Surface Vehicles (ASVs) is presented with extensive path planning and abstraction capabilities of the environment. The aim is to enable the ASV to pass through heavy harbour traffic to an intended destination safely and efficiently. While predicting the future path of other vehicles and path planning can be carried out by known methods, a remaining challenge is to guarantee that all planning computations are completed in a timely manner. Building on considerable effort in the literature on how to steer an ASV in a dynamical environment, where a number of vehicles operate simultaneously, this paper defines environmental complexity and a logical framework for controlling the boat which takes into account computational performance. Implementation techniques of this theory are illustrated on agent-based mission-management of an autonomous surface boat for mine detection and disposal.