Localization Tracking for Autonomous Surface Vehicles Using You Only Look Once version 5 in ROS Gazebo Simulator

Abstract

Autonomous Surface Vehicles (ASVs) have emerged as a significant research focus due to their extensive applications. A major challenge in ASV development is the rapid and accurate detection and identification of objects on the water's surface, such as buoys. This study investigates the use of YOLOv5 for buoy detection on ASVs, focusing on path localization within the Robot Operating System (ROS) framework. Path localization is employed to determine the vessel's route and its movement through the test path based on buoy detection. The results indicate that YOLOv5 achieved a 100% accuracy rate in detecting boundary buoys within the Gazebo simulator, a component of the ROS ecosystem. Furthermore, the ASV was able to navigate precisely along the center of the test path without colliding with the boundary buoys. This study is expected to make a significant contribution to the advancement of ASV technology.