The integration of artificial intelligence (AI) into unmanned aerial vehicles (UAVs) enables autonomous, smart surveillance and rapid disaster response. AI-enabled drones combine high-resolution vision, thermal imaging, and advanced decision algorithms to detect, track, and assist in emergencies with minimal human intervention. We propose a modular system architecture comprising visual and infrared perception, navigation control, 5G/IoT communication, and onboard AI decision-making (e.g. YOLOv8 object detection, Open Pose estimation). The implementation uses a multirotor drone equipped with RGB and thermal cameras, an edge AI accelerator (e.g. NVIDIA Jetson), and custom control software (PID flight control, path planning). In experiments, the system achieved real-time detection (YOLOv8 at ~50 FPS on Jetson) with high precision (>95%) and significantly reduced response time in SAR simulation. A case study of a mountain rescue scenario demonstrates dramatically faster victim localization via thermal YOLO tracking. This AI-augmented UAV platform offers improved coverage and efficiency over traditional methods, promising lower costs and higher safety for disaster management.
You may also start an advanced similarity search for this article.