Robust Dual-Camera Lane-Keeping System Using YOLOv8 and ROS2 for Autonomous Driving Education

초록

This paper presents a study on the development of a lane-keeping system using front and rear dual cameras, a core technology for advancing AI-based autonomous driving education. To expand traditional autonomous driving education, which focuses on static obstacles, to a more realistic dynamic environment: stable driving control technology must be established first. To this end, this study proposes a system that recognizes lanes in real-time using the one-stage object detection model YOLOv8 and generates and follows a driving path based on this recognition. The entire system is based on the Robot Operating System (ROS2), with perception, path planning, and motion planning modules designed as independent nodes to ensure flexibility and scalability. The proposed system provides a stable foundation for adding complex scenarios such as dynamic obstacle avoidance and traffic light recognition in the future. Its performance is verified through driving tests in a closed-loop track environment.

키워드