Research on Autonomous Landing Method of Shipborne Uncrewed Aerial Vehicle Based on Visual Recognition

To address the challenges of GPS signal failure, dynamic vessel motions, and low visual recognition confidence during shipborne UAV maritime landings, this study proposes an integrated closed-loop solution harmonizing strategy, algorithm, and hardware components, incorporating a staged landing approach, an enhanced YOLOv8 object detection algorithm with C2f-RVB module, ABF feature fusion module, and dedicated small-target detection layer, along with a KCF-based re-detection mechanism and hardware innovations featuring multi-ArUco composite markers and a self-clamping landing platform, which collectively enhance autonomous landing capabilities in complex maritime environments, where experimental results demonstrate a mAP0.5 of 93.5% for the modified YOLOv8 and 10% to 15% higher detection confidence under non-horizontal or partially occluded marker conditions, thereby establishing a technically robust framework balancing precision and operational robustness for shipborne UAV autonomous landings.