船舶自動航行システムにおける複数船認識とその学習

概要

論文の詳細を見る
Strategy for multi-ship encounter problem is not yet established in automatic collision avoidance system. In SAFFS (Ship Auto-navigation Fuzzy Expert System) proposed by the first author and others, it is treated to avoid one-by-one for the most dangerous ship among other ships each reasoning moment. Some dangerous or wasteful manoeuvre may occur at certain occasions. To improve this a new concept of avoiding a group of ships is proposed. The principle ideas of defining "group" and strategies for avoiding groups are dealt with. On the other hand, in SAFES, there is no feature of improving the rules, degree and timing of avoiding. This may be called the learning feature. As an idea of this feature, a simple random-walk optimization is proposed and validated by simulation. Although as a goal of ship automatic navigation system, both features should be included to compensate each other, as a first step, they are independently treated in this paper.
社団法人日本船舶海洋工学会の論文
1996-09-25