YOLOv11-Powered Emergency Vehicle Detection Algorithm Using Tello Quadrotor Drone
Alicia Loh I-Ling, Goh Yu Xiang, Kishore Bingi, Rosdiazli Ibrahim · 2025
This study presents a YOLO-powered vehicle detection and tracking system integrated with a Tello quadrotor drone, addressing the limitations of conventional detection systems. Traditional methods often generalize vehicles into a single category, overlooking their specific operational needs and urgency levels. Furthermore, stationary and GPS-based tracking systems face constraints in coverage, environmental adaptability, and signal clarity. Leveraging the agility and real-time capabilities of the Tello drone, combined with the YOLOv11 segmentation model, this research introduces a robust framework for differentiating emergency vehicles such as ambulances and fire trucks. The system employs a carefully constructed dataset of annotated images processed using Roboflow and YOLOv11. The model achieves high accuracy metrics, including mean average precision (mAP) scores above 0.95 for bounding boxes. Real-time deployment demonstrates the model's reliability in detecting emergency vehicles under diverse scenarios, with ambulances exhibiting slightly better detection consistency than fire trucks due to their standardized design. Challenges such as variable lighting and occlusions were noted but addressed through data augmentation and hyperparameter optimization. This research highlights the potential of drone-based AI solutions to enhance emergency response, traffic management, and public safety, setting a foundation for future advancements in dynamic, real-world applications.