다중선박 충돌회피 알고리즘에 관한 연구

Abstract

Auto-navigation algorithms have been studied to prevent maritime accidents such as collision and grounding due to human error and carry out navigation safely and effectively. However, because most of collision avoidance algorithms have never been validated on the real coastal traffic situation together with many traffic ships, they are not successfully applied to the real ship. In this study, in order to support safe navigation and reduce the probability of collision and grounding, a Collision Avoidance algorithm is developed by using Fuzzy algorithm and the concept of Changeable Action Space Searching (CAS). In the first step, on a basis of collision risk calculated from fuzzy algorithm in the current time(t=to), alternative Action Space for collision avoidance is planned. In the second step, next alternative Action Space for collision avoidance in the future(t=to+Δt) is corrected and re-planned considering changed collision risk. In the third step, the safest and most effective course out of Action Space is decided by using optimization method in real time. In this paper, the main features of the developed collision avoidance algorithm (CAS) are introduced. CAS is implemented in the ship-handling simulator of MOERI. The performance of CAS is tested on the situation of open sea with 3 traffic ships, whose position information of AIS is assumed and used. Own-ship is fully autonomously navigated by auto-pilot for the course of collision avoidance, which is calculated from CAS. As a result of experiments, it is found that the calculated courses for collision avoidance from CAS are reasonable and own-ship can successfully avoid the collision against traffic ships.