TACK
TACK - Tunnel Automatic CracK Monitoring using Deep Learning
Ongoing research
The ongoing research focuses on moving one step closer towards implementing a digital inspection technique for tunnels. The aim is to work closely with inspectors and owners of infrastructures to validate and show the reliability of the results from an autonomous and digital inspection.
To accomplish this, in the next year the research activities will focus on the following WP:
WP1 Data collection and management
collection of data in 2-3 tunnels in collaboration with WSP and manual inspectors
WP2: Crack detection and measurement
improvement of the deep learning algorithm for automatic crack detection
investigation of trained model transferability
WP3: Data visualization
development of a methodology to present the results from the digital inspection in a user-friendly visualization tool
WP4: Structural assessment of damaged tunnels
investigation of risk assessment associated with cracks in tunnel linings
definition of threshold values for different risk levels of cracks
WP5: Comparison of digital and human inspection
evaluation and comparison of the results from a digital inspection to in-situ inspections performed by experts