Collision avoidance of maritime vessels

Wasif Naeem, Sable Campbell, Mamun Abu-tair

Research output: Chapter in Book/Report/Conference proceedingChapter (peer-reviewed)


This chapter presents an artificial potential field (APF)-based online collision avoidance system for manned and unmanned maritime vehicles, which is capable of reacting to static and dynamic obstacles in the vicinity. The standard marine `rules of the road' are integrated into the collision avoidance framework. A risk assessment module is also introduced which is based on the standard closest point of approach (CPA) method. A decision maker then selects appropriate rules based on relative heading and positions of the vessels. For the detection part, an integrated vision and laser-based system has been developed to provide sensing functionality for multiple obstacles. Autonomous craft are preferred over manned vessels in scientific operations as they are more suited for long enduring and tedious missions in dangerous or hazardous environments. The technique presented is fairly generic and is applicable to a general class of marine vehicles ranging from a small/medium-sized craft to a large freighter or an oil tanker. The small/medium-sized vehicles have been widely employed in surveillance and scientific missions. For surveillance missions, deploying autonomous vehicles will maximise the coverage and reduce the number of personnel involved in the operation area. Simulation results are provided to include the three fundamental collision encounter scenarios, that is, overtaking, head-on and crossing. It is also shown that Dubin circles could be successfully employed to take the dynamics of the craft into account, which provides a general method independent of the craft size.
Original languageEnglish
Title of host publicationNavigation and Control of Autonomous Marine Vehicles
Number of pages24
ISBN (Electronic)9781785613395
Publication statusPublished - 2019


Dive into the research topics of 'Collision avoidance of maritime vessels'. Together they form a unique fingerprint.

Cite this