A Hybrid-Type Power Transmission Line Inspection Drone and an Anomaly Detection Method Using a Deep Neural Network

A hybrid-type power transmission line inspection drone and an anomaly detection method using a deep neural network are developed in this study. Since power transmission lines are one of the crucial infrastructures for our society, their regular maintenance and damage assessment are required. Manual inspections are currently performed by human workers who climb up transmission towers and navigate through the lines, which is time-consuming and poses significant risks. The use of drones or robots can improve efficiency and reduce the risks associated with manual inspections. However, drones need multiple flights to observe the entire line and expensive laser sensors, and robots still require human workers to climb up the transmission tower. The hybrid-type drone developed in this study takes off from the ground and lands on power transmission lines, then navigates along them like a conventional robot. The drone has a footprint and height of 100 cm × 120 cm and 80 cm, respectively, weighs about 11 kg, and is equipped with four propellers for flying and two motor-driven wheels for navigating along power transmission lines. The drone captures high-resolution images of the lines using three cameras mounted on it, and an artificial intelligence-based inspection system has also been developed to detect damages or anomalies in the images. The system employs autoencoders based on convolutional neural networks for image processing and anomaly detection.