Applying Automation and Machine Learning for Tunnel Inspections - RETC2023

Tunnel inspections have traditionally been carried out manually by inspectors writing up observations and taking photos of defects. The results from the inspection and the defects observed are dependent upon the rigor of the inspectors and may be subject to repeatability and consistency issues and is often time consuming with elevated health and safety risks. This paper will discuss work that is being carried out in developing and implementing an innovative hardware and software solution integrating machine learning to automate the process of capturing objective tunnel condition information, offering cost and programme savings as well as health and safety improvements