TACk - Tunnels Automatic CracK Monitoring using Deep Learning
| Reference number | |
| Coordinator | Kungliga Tekniska Högskolan - Urban Planning & Environment |
| Funding from Vinnova | SEK 2 174 236 |
| Project duration | June 2019 - October 2022 |
| Status | Completed |
| Venture | The strategic innovation program InfraSweden |
| Call | Solutions for sustainable transport infrastructure |
Important results from the project
The aim of this project was to create an autonomous inspection and assessment system of tunnels based on the combination of image processing techniques and deep learning algorithms. To achieve this, the aim was to raise the overall TRL (Technology Readiness Level) of the proposed system from the TRL 3/4 (TRL3 “experimental proof of concept” TRL4 “technology validated in a laboratory”) to a TRL7 system prototype demonstration in an operational environment.
Expected long term effects
We achieved promising results for crack detection and measurements combining CNN and photogrammetry. The detailed mapping of cracks and the possibility to measure their width gives a highly efficient basis to assess the need for maintenance of tunnels. This will improve the knowledge regarding tunnel conditions and facilitate maintenance planning which, hopefully, will reduce tunnel downtime and the cost of monitoring and maintenance. Moreover, digital twins will facilitate knowledge transfer between inspectors and owners to improve decision-making.
Approach and implementation
The project was carried out as a collaboration between KTH, Sapienza University of Rome, and WSP Sweden. Algorithms for crack detection and measurements were implemented and validated. The achieved results were presented at scientific conferences, as master thesis projects, and as papers in high-ranked scientific journals. Also, datasets from laboratory tests will be soon published online together with the tunnel crack labeled dataset to help the scientific community.