A Comparative Study of Vision-Based Detect and Avoid for Urban Air Mobility

Urban Air Mobility (UAM) aircraft require reliable methods to avoid collisions at low altitudes, where radar-based solutions may be limited or costly. The paper explores a vision-based Detect and Avoid (DAA) method using three cutting-edge detection models: YOLOv9, YOLOv11, and Detection Transformer (DETR). All models were trained on the AVOIDDS dataset, which contains over 72,000 images of intruder aircraft under various times of day, weather conditions, geographic regions, and aircraft types. Compared to YOLOv8, our results show that DETR achieves the highest accuracy at 94.2% mAP, while YOLOv9 provides 2.5× faster processing speed with 89.1% mAP. YOLOv11 achieves 88.5% mAP, making it suitable for electric vertical take-off and landing (eVTOL) and unmanned aircraft systems (UAS) with limited computational resources. These findings highlight the trade-off between processing speed and detection accuracy, emphasizing that model selection should be guided by each aircraft’s operational constraints. By leveraging these advanced detection models, we aim to improve airspace operational safety.