무인선의 충돌회피 시스템 검증에 관한 연구

Abstract

In an unmanned surface vehicle (USV) for ocean observation and sea surveillance, it is necessary to develop the collision avoidance system against traffic ships and obstacles. In this paper, a collision avoidance system (CAS) of USV is developed by using changeable action space searching. In order to evaluate the performance of CAS, a jet-boat is used as test-bed. USV avoids against obstacles such as fixed shorelines and moving traffic ships. Navigational information of USV is collected by using RTK-DGPS and gyro. Obstacle can be recognized by using AIS. A changeable action space is searched by using fuzzy inference on collision risk. The alternative route for collision avoidance is optimized according to the cost function of collision risk. In order to evaluate the performance of CAS for USV, field test are carried out on head-on and crossing. In this paper, the main features of CAS and main results of field test are described.loped by using changeable action space searching. In order to evaluate the performance of CAS, a jet-boat is used as test-bed. USV avoids against obstacles such as fixed shorelines and moving traffic ships. Navigational information of USV is collected by using RTK-DGPS and gyro. Obstacle can be recognized by using AIS. A changeable action space is searched by using fuzzy inference on collision risk. The alternative route for collision avoidance is optimized according to the cost function of collision risk. In order to evaluate the performance of CAS for USV, field test are carried out on head-on and crossing. In this paper, the main features of CAS and main results of field test are described.