Study of artificial intelligence and computer vision methods for tracking transmission lines with the AID of UAVs

Abstract

Currently, Unmanned Aerial Vehicles (UAVs) have been used in the most diverse applications in both the civil and military sectors. In the civil sector, aerial inspection services have been gaining a lot of attention, especially in the case of inspections of high voltage electrical systems transmission lines. This type of inspection involves a helicopter carrying three or more people (technicians, pilot, etc.) flying over the transmission line along its entire length which is a dangerous service especially due to the proximity of the transmission line and possible environmental conditions (wind gusts, for example). In this context, the use of UAVs has shown considerable interest due to their low cost and safety for transmission line inspection technicians. This work presents research results related to the application of UAVs for transmission lines inspection, autonomously, allowing the identification of invasions of the transmission line area as well as possible defects in components (cables, insulators, connection, etc.) through the use of Convolutional Neural Networks (CNN) for fault detection and identification. This thesis proposes the development of an autonomous system to track power transmission lines using UAVs efficiently and with low implementation and operation costs, based exclusively on rea-time image processing that identifies the structure of the towers and transmission lines durin the flight and controls the aircraft´s movements, guiding it along the closest possible path. A sumary of the work developed will be presented in the next sections.