Ship Collision Avoidance by Distributed Tabu Search

Abstract

More than 90% of world trade is transported by sea. The size and speed of ships is rapidly increasing in order to boost economic efficiency. If ships collide, the damage and cost can be astronomical. It is very difficult for officers to ascertain routes that will avoid collisions, especially when multiple ships travel the same waters. There are several ways to prevent ship collisions, such as lookouts, radar, and VHF radio. More advanced methodologies, such as ship domain, fuzzy theory, and genetic algorithm, have been proposed. These methods work well in one-on-one situations, but are more difficult to apply in multiple-ship situations. Therefore, the authors proposed the Distributed Local Search Algorithm (DLSA) to avoid ship collisions as a precedent study. DLSA is a distributed algorithm in which multiple ships communicate with each other within a certain area. DLSA computes collision risk based on the information received from neighboring ships. However, DLSA suffers from Quasi-Local Minimum (QLM), which prevents a ship from changing course even when a collision risk arises. In their study, the authors developed the Distributed Tabu Search Algorithm (DTSA). DTSA uses a tabu list to escape from QLM that also exploits a modified cost function and enlarged domain of next-intended courses to increase its efficiency.They conducted experiments to compare the performance of DLSA and DTSA. The results showed that DTSA outperformed DLSA.