On the sea trial test of the autonomous collision avoidance among multiple unmanned surface vehicles

Abstract

International Maritime Organization (IMO) adopted the term of Maritime Autonomous Surface Ships (MASS) as a concept of future ship at the Maritime Safety Committee (MSC) in 2017. Unmanned Surface Vehicle (USV) has been developed in order to perform difficult missions as a substitute of manned vehicle in bad sea weather condition. USV is a small boat as a kind of small MASS and it can be controlled remotely or autonomously. Since 2011, Korea Research Institute of Ships and Ocean Engineering (KRISO) have developed ARAGON as a multi-purpose intelligent unmanned surface vehicle (USV) under the financial support of KOREA Ministry of Oceans and Fisheries. Three prototypes USV ARAGON1, ARAGON2, ARAGON3 have been built and they have planning hulls with diesel engine and waterjet. They have the length of 8 meters and the displacement of about 3 tons. On the other hand, because USV has various merits of higher payloads and longer operation time in sea area than aerial drone and multiple USVs are more effective for maritime mission such as intelligence, surveillance, reconnaissance than single USV, a lot of research and development projects on USV swarm have been conducted. In KRISO, a new research project entitled with Development of situation awareness and autonomous navigation technology of unmanned surface vehicle based on the artificial intelligencehas been studied for the development of the operation system of USV swarm for illegal ship control support and sea surveillance since 2019. For this, swarm autonomous navigation system (SANS) for remote control, path following, collision avoidance and illegal ship chasing has been developed by using four USVs, which are ARAGON1, ARAGON2, ARAGON3 and KOMBO. For safe navigation among four USVs swarm, they should keep the convention on the international regulations for preventing collisions at sea, 1972 (COLREGs). Collision avoidance algorithm in SANS is implemented by using changeable action space searching, which can be flexibly changed according to the collision risk (CR) based on fuzzy inference. Each USV can navigate autonomously without human operation by using SANS. In the ground control station the navigational information and the status of four USVs are remotely monitored through LTE communication. In the lake, field tests are carried out in order to validate the collision avoidance algorithm among four USVs swarm. Scenarios are tested on the simultaneous colliding situations such as head-on, crossing and their combinations among four USVs swarm. In this paper, the main features of SANS and main results of field tests on autonomous collision avoidance among four USVs swarm are described. ？ 2023 IEEE.