Automated Inspection and Fault Detection in Rooftop Solar PV System Using YOLO-Driven Tello Quadrotor Drone

This study presents an AI-driven drone inspection system for rooftop solar PV panels, employing the DJI Mavic Mini drone and YOLOv11 AI model to streamline defect detection. The methodology involved capturing high-resolution images via autonomous drone missions, annotating data with RoboFlow, and training the YOLOv11 model to accurately identify defects such as cracks, discolouration, and delamination. The system achieved a mean Average Precision (mAP) of 0.577 at an IoU threshold of 0.5, validating its accuracy in real-world scenarios. A significant innovation was the creation of a user-friendly application automating orthomosaic image generation, segmentation, defect detection, and reassembly of annotated maps, minimizing manual effort and errors. Results demonstrated efficient localization and classification of defects across the panels, though challenges persisted with subtle defect types. Future efforts will focus on refining detection algorithms, expanding datasets, and scaling the system for large-scale solar farms to enhance operational scalability and sustainability.