Development of Automated Traveling Control for Transfer Cranes Using GNSS and White Line Image Recognition

In recent years, the introduction of remote and autonomous driving technologies has advanced in port container logistics to improve working conditions, stabilize operations, and enhance safety. Among them, the automation of transfer cranes that load and unload containers in storage yards has become a key element in realizing smart ports. RTG(Rubber Tired Gantry Crane), widely used in Asia and Japan for their flexibility, require highly accurate travel and stopping control since they operate without rails. Traditionally, straight-line driving control with high-precision RTK-GNSS (Real Time Kinematic-Global Navigation Satellite System) has been applied, but as remote automation progresses, high stopping accuracy has become equally essential. However, GNSS sensors mounted on top of RTGs are affected by pitching due to slopes or acceleration, leading to discrepancies between measured and actual positions. To address this, researchers have explored the use of deep learning-based image recognition techniques to detect existing white lines in container yards, though challenges remain under severe outdoor conditions such as rain or glare. This study combines GNSS with YOLOX-based robust real-time white-line recognition to achieve accurate travel and positioning. A control system incorporating Input Shaping, Kalman Filtering, and feedback control was implemented to ensure seamless switching between GNSS and vision-based positioning, suppress pitching, and enhance stopping accuracy. Experiments on actual RTGs confirmed the effectiveness of the proposed system.