Safe path planning for autonomous ships based on the point of potential collision concept

Abstract

The development of Autonomous Navigation Systems (ANS) constitutes a milestone in the process leading to the achievement of the autonomous operation of different vehicles, such as cars, aircrafts, mobile robots and ships. Obstacle avoidance is a vital task that has to be included in the process of the development of such system. The paper introduces a path planning method to be applied in the Maritime Intelligent Transportation Systems (M-ITS). The method uses the safety indicators such as the Distance at the Closest Point of Approach (DCPA), the Time to the Closest Point of Approach (TCPA), the Bow Crossing Range (BCR) and the Bow Crossing Time (BCT). The method considers the International Regulations for Preventing Collisions at Sea (COLREGs) and safety zones (domains) around target ships. The deterministic nature of the algorithm guarantees the repeatability of solutions for every run of the algorithm with the same input data and very short run-time. The approach was compared with two other algorithms, one deterministic and one heuristic. Results of 100 test cases with different complexity (simple encounters defined in the COLREGs as well as multi-ship encounters with up to 20 target ships) were compared in the paper. The algorithm searches for a solution in a cascade manner, what allows for the achievement of competitive results as compared to other algorithms maintaining short run time. These features allow to apply this collision avoidance (COLAV) method in commercial systems of autonomous ships and Unmanned Surface Vessels (USVs).