R-YOLO: Enhancing Takeoff/Landing Safety in UAM Vertiports With Deep Learning Model

This study introduces the R-YOLO model, an innovative AI-driven safety system that incorporates real-time object detection and trajectory prediction to provide decision support information during the critical takeoff and landing phases of improving urban air mobility (UAM) operations. The proposed unified approach, in contrast with traditional techniques, effectively combines detection and prediction into a single model, allowing effective real-time video processing. Leveraging data from onboard cameras, the model continuously monitors vertiport surroundings, enabling instantaneous situational awareness and predictive risk assessment. This method offers useful information about object positions and movement, which improves situational awareness, reduces operator workload, and facilitates better decision-making. The primary contribution of the research is the unified structure for anticipatory safety mechanisms, a step forward in AI-enabled aviation. By addressing major issues in urban aviation with this innovative artificial intelligence framework, the study provides the possibility of safer and more effective UAM operations.