Automatic Collision Avoidance System for Ships Using Fuzzy Control

Abstract

For the ultra automatic ship operating system, collision avoidance system is one of the most difficult problems. Usually highly experienced experts such as captains etc. can make a suitable decision considering various environmental conditions. The authors propose an automatic collision avoidance system controlled by an on-board computer. The system contains of the following four parts; data acquisition from radar signal, decision of the collision risk, speed or course change command and course control autopilot. All parts except data acquisition are done by fuzzy reasoning or fuzzy control. Fuzzy control is quite similar to the human control and so flexible even to unexpected environmental changes. First, system design is done and algorithms of each sub-system are described. It is not necessary to apply fuzzy control for course keeping and course changing, but we dare to apply it. Fuzzy control autopilot is very smooth and at the same time very quick for an abrupt course change. Computer simulation is done for the various relative conditions of a target ship and the results are found to be reasonable and reliable as compared with human behaviour carried out at out ship handling simulator. At last, the system is also applied to the collision avoidance very close to the course changing point or with conflicting behaviours of a target ship. For both cases, the system is found to be worked properly.