실환경에서의 USV-UAV간 이종협응제어 연구

Abstract

This paper presents a study on the development of a cooperative control systembetween Unmanned Aerial Vehicle(UAV) and Unmanned Surface Vehicle (USV) in a real-world environment. The proposed system receives real-time aerial imagery from UAV, identifies obstacles and water tanks, and generates paths for USV to follow. The system was implemented using the Robot Operating System(ROS), and Domain Bridge was utilized to address potential data latency issues when operating across heterogeneous platforms, allowing the transmission of only essential data between platforms to minimize latency. Real-time aerial imagery was processed using the YOLO(You Only Look Once) deep learning model to detect obstacles, basin, and ships with high accuracy and efficiency. The detected data was used to generate optimal paths through the A* global path planning algorithm. To ensure accurate path tracking, a guidance algorithm based on LOS with CA was developed, and the generated paths were followed using a PD control method to maintain stability and precision. Before conducting real-world tests, the system's stability was verified in a rendering-based simulation environment(Gazebo). Subsequently, field experiments were conducted under the same conditions to validate the system's performance. A comparison with the DWA (Dynamic Window Approach) local path planning algorithm was also performed, demonstrating that the proposed system showed higher efficiency in energy consumption, total distance traveled, and mission completion time. This study provides insightful contributions to various maritime scenarios and proposes a new direction for the integrated operation of UAV and USV in maritime environments.