Empowering automated surface defect inspection with IoT and deep learning techniques: an application in Trunk Road T2 and Cha Kwo Ling Tunnel, Hong Kong

The continuous collection and analysis of data are instrumental in the inspection of tunnel structures. This procedure enables tunnel inspectorates to systematically monitor structural integrity, thereby ensuring the safety and stability of tunnels. Thanks to recent breakthroughs in technologies such as Internet of Things (IoT), deep learning, and robotics, it is now feasible to automate the inspection of tunnel structures. Hong Kong Productivity Council (HKPC), Civil Engineering and Development Department (CEDD), and Hyder-Meinhardt Joint Venture (HMJV) have collaboratively developed an AI-based tunnel structure inspection system that integrates air-ground cooperation. This system leverages cutting-edge techniques, including IoT sensors, Unmanned Ground Vehicle (UGV), and Unmanned Aerial Vehicle (UAV), to facilitate the real-time image data capturing of tunnel structures. Furthermore, it is equipped with the capability to automatically identify defects such as concrete cracks, concrete spalling, and water leakages. Consequently, this advancement not only heightens the efficiency of the tunnel inspection process by diminishing the inspection time and increasing cost savings, but also enhances the site safety and optimises the construction logistics.