Evasive Decision Making in Uninhabited Maritime Vehicles

Abstract

AbstractUnmanned surface vehicles (USVs) are able to navigate in hazardous environments without jeopardising human life. In order to eliminate the human factor altogether, a reliable onboard collision detection and avoidance system is mandatory. This would improve a vessel's self reliance and aid in accomplishing complex missions for extended periods. In addition, the evasive manoeuvres are required to be compliant with coastguard regulations (COLREGs) defined by the International Maritime Organisation. This paper reports the preliminary investigation of such collision avoidance manoeuvres by employing a biased waypoint guidance scheme and genetic algorithms. The paper deals with reactive collision avoidance for objects in the near field. In addition, the dynamics of a USV called Springer is incorporated which provides a realistic setting for the proposed strategies. Simulation results demonstrate that the adopted schemes are able to generate safe and viable reference trajectories.