표식 인식 기반 무인수상체의 자동 도킹시스템 개발

Abstract

In this study, autonomous docking software and hardware system for ASV(Autonomous Surface Vehicle) based on symbol recognition are developed. A symbol recognition based docking guidance necessarily includes a step for symbol recognition. Vision sensors such as a camera and distance measuring sensors like LiDAR(Light Detection And Ranging) usually used for recognition. Due to the feature that requires various sensors, the performance of the environment recognition-based docking algorithm should be verified through actual tests rather than simulations in a virtual environment. Therefore, ROS(Robot Operating System) is used as an operating system to develop an autonomous system. ROS makes it possible to communicate between sensors and operate actuators. In terms of algorithm composition, it is divided into guidance, navigation, and control nodes based on the GNC(Guidance-Navigation-Control) theory. This system makes it possible to perform remote control, waypoint tracking, and docking tasks respectively, according to the situation. The docking guidance proposed in this study is an environment recognition-based that allows the ASV to be guided to the docking port by recognizing the surrounding environment including the symbols without using GNSS(Global Navigation Satellite System) data. Therefore, the docking process is divided into an approach phase and a docking phase. In the approach phase, LOS(Line-Of-Sight) guidance is used for approaching near the docking port. Docking phase consists of an image processing algorithm for recognizing a symbol and a docking algorithm leading ASV to a designated docking port. The desired heading of the docking algorithm is determined by the combination of the centering angle and lead angle: centering angle for pure tracking and lead angle for guiding the ASV to the safety zone. The centering angle and the lead angle are determined according to the heading and position of the ASV. To estimate the relative position of the ASV from the dock, the information of the dock and the symbol detected by the camera and LiDAR is used. When the ASV is in a posture that can not recognize the symbol, the ASV searches the surroundings to find the symbols through a searching algorithm. To verify the docking ability of the proposed docking algorithm, on-land and tank docking tests are implemented. The performance of the developed autonomous docking system was verified by participating in the docking mission of KABOAT 2021(Korea Autonomous BOAT 2021).