Runway Multi-functional Robot and Small Targets Detecting Method

Foreign Object Debris (FOD) is the second most significant aviation safety hazard after bird strikes. Existing FOD detection equipment, situated at a considerable distance from the runway, often results in poor imaging quality for small targets, which impacts detection performance and is also costly. Therefore, this study has developed a highly mobile and low-cost multi-functional runway operation robot that can conduct rapid automatic inspections during runway idle periods. By equipping the robot with high-resolution cameras and computing devices, the detection accuracy and real-time performance of FOD can be significantly improved. To further enhance the detection effectiveness of FOD, this study also proposes a FOD data augmentation method and an improved network structure based on YOLO. Experimental results demonstrate that the refined model has achieved a notable improvement of $\mathbf{2 . 1 \%}$ in precision compared to the traditional YOLOv5 model, with a frame rate increase of approximately 45 frames.