Decision-making algorithm for ship collision avoidance with collision risk map

• Published in: Ocean engineering 2023-10, Vol.286, p.115705, Article 115705
• Main Authors: Yoshioka, Hitoshi, Hashimoto, Hirotada, Makino, Hidenari
• Format: Article
• Language: English
• Subjects: Collision avoidance; Collision risk map; Decision-making; Optimal route-planning
• ISSN: 0029-8018; 1873-5258
• DOI: 10.1016/j.oceaneng.2023.115705

This paper presents a decision-making algorithm for ship collision avoidance. First, an algorithm for the planar distribution of collision risk is described. The level of collision risk was calculated as a function of the plane coordinates based on the predicted encounter state. It enables the visualization of all the potential collision risks as a collision risk map and the calculation of the integrative collision risk against several ships, considering COLREGs. Second, an optimal route-planning algorithm that can consider the maneuvering characteristics of the ship is proposed. Because a ship cannot change its course and speed suddenly, the nodes and their connections can be set to limit the course and speed change amounts to appropriate ranges. The optimal route was obtained as the minimum-cost route along node connections. Finally, a decision-making algorithm for ship collision avoidance was developed by combining the proposed collision risk mapping and optimal route-planning algorithm. It was applied to representative encounter situations and evaluated by an evaluation method. Further, the proposed algorithm was compared with state-of-art algorithms. Thus, safe and efficient decision-making for collision avoidance can be achieved even in congested situations by appropriately changing both a course and a speed depending on the encounter situation. •An algorithm to calculate a planar distribution of collision risk considering COLREGs is proposed.•An algorithm for optimal routing for collision avoidance by both course and speed changes is proposed.•A decision-making method for collision avoidance is developed by combining the proposed two algorithms.•The developed method can provide a safer and more efficient collision avoidance plan even in congested situations.