Abstract: unmanned aerial vehicle drones system for insulator cleaning. The system utilizes the global positioning system to guide the UAV in automatically searching for insulators and returning to the charging station. A customized nozzle，robotic arm，and water pump are integrated to form the cleaning tool. During the automatic cleaning phase，the YOLOv4 Tiny+TensorRT object detection model is employed to detect and track insulators，while a fully convolutional network segments the dirty areas to identify optimal cleaning points. For autonomous landing，a position-based visual servo control method is adopted. Through 30 cleaning and landing experiments，the reliability of the automatic cleaning and the safety of autonomous landing were validated. The results show that the dirty area on insulators was reduced by approximately 70% on average after cleaning，and the landing error relative to the center of the charging platform was less than 12 cm.