Drone-Based Corrosion Detection on High-Voltage Transmission Towers Using Hyperspectral Imaging

High-voltage transmission towers require regular inspections to identify corrosion. Traditionally, these inspections are performed through climbing, involving skilled technicians. This method is both tedious and hazardous, often necessitating the shut-down of sections of the high-voltage grid. In this paper we propose a workflow that relies on drone-based hyperspectral imaging, which enables remote assessment without endangering the technician. Currently, drones are equipped with conventional RGB cameras. However, these cameras have limited spectral resolution and range, which compromises their ability to reliably detect corrosion and often leads to false alarms. Moreover, conventional RGB cameras are unsuitable for accurately assessing the severity of corroded areas. To address these challenges, this study proposes a solution that leverages hyperspectral imaging and a dedicated processing pipeline to robustly detect corrosion and classify it based on severity level. Experiments using drones equipped with imec’s VIS-NIR hyperspectral payload demonstrated the effectiveness of our developed solution.