Automatic Collision Avoiding Control of Own Ship in Critical Cases Based on Game Theory

摘要

Ensuring safety of navigation is always a matter of great importance for the ship officers. Thank to the advanced on-board equipments, ship officers now have the access to a huge amount of environment and targets information which can be used to determine a collision avoiding strategy for the own ship. However, maritime accidents are still happening, mostly due to human errors in judging information or controlling the ship, especially in critical cases where he is overwhelmed by the imminent risks of collision. In critical cases, the time allowed for decision making is extremely small and the intention of target ship is unpredictable. The nature of collision avoiding maneuver is therefore similar to the nature of a pursuit-evasion game in which the own ship plays the role of an evader trying to avoid the collision while the target ship is a pursuer who is aiming at causing it. Thence, in this study, the game theory is used to model the collision avoiding problem between the ships as a pursuit-evasion game, taking into consideration the own ship maneuverability. Bacteria foraging optimization algorithm is then applied to seek an approximation of the optimal solution to the problem. Later on, simulation is carried out with an MMG model for the own ship and target motions perceived by radar in Tokyo Bay to verify the method viability.