On an Autonomous Navigation System for Collision Avoidance of Unmanned Surface Vehicle

An autonomous unmanned surface vehicle (AUSV) has been developed for multipurpose mission such as ocean observation and sea surveillance. The first prototype AUSV has been designed to have the length of about 7 meter and it has single water-jet with diesel engine. An autonomous navigation system is designed for collision avoidance of AUSV against obstacles such as fixed shorelines and moving traffic ships. Navigational information of AUSV is acquired by using Real-time Kinematic (RTK) DGPS and Fiber-optic gyro (FOG). Automatic Identification System (AIS) is also used to recognize obstacle. In this paper, an action space searching algorithm for collision avoidance (ACA) is newly designed by using fuzzy algorithm. The nodes and layers for searching route for collision avoidance are real-timely arranged around AUSV. In each node, collision risk between AUSV and obstacles is real-timely calculated by using fuzzy algorithm. Collision risk (CR) is an index, which is derived from DCPA and TCPA with obstacles through the inference of fuzzy algorithm. Here, main dimensions and velocity of AUSV are also considered. The paths, which are connected from current position of AUSV, node and waypoint are alternative routes for collision avoidance. The optimal path for collision avoidance is real-timely decided considering cost function with collision risk (CR), the time integration of (CR) or the integration of DCPA. Here, International Regulations for Preventing Collisions at Sea (COLREGs) are also considered. Auto-pilot controls nozzle actuator of water-jet according to the optimal path. In order to evaluate the performance of newly-designed ACA, numerical simulation and field test are carried out on several scenarios of significant situation such as head-on, crossing, and over-taking and multiple ships' colliding. In this paper, the main features of newly designed action space searching algorithm for collision avoidance (ACA) and main results of numerical simulation and field tests are compared and introduced.