Project Summary

The project aims to showcase the potential of machine learning and image processing in detecting and tracking defects in brickwork transport infrastructure. The scope covers capturing image data through remote platforms, creating precise 3D photogrammetric models, and generating a spatial 3D database of identified defects. The resulting database can be utilised by engineers to evaluate asset deterioration.

Project Achievements

Next Steps and Future Measures of Success • The further development of a survey vehicle to collect high resolution survey data in tunnels. • The production of detailed 3D models of assets using photographic data collected. • The processing of photographic data using AI to identify issues in brickwork. • Updated algorithm to display photographic data and AI results on 2D engineering drawings.

Conclusions

The project demonstrates conclusively that AssetScan can be used in the context of transport infrastructure, particularly assets where the structural fabric is brickwork. AssetScan could therefore be used in combination with asset owner data collection programmes to automatically highlight defects and bring these to the attention of engineers quickly. With the demonstrated orthorectification techniques, this could be undertaken on a regular basis and used for automated change detection.

Next Steps

• CCI will seek early adopters in the transport sector. • Split funded commercial projects and additional grant funding for development. • Further data science and product development towards TRL6. • Development of outputs towards a BIM environment. • Additional data introduced into training datasets with manual labels. • Validation on a project-by-project basis increasing to 95% accuracy. • Commercial growth of CCI including procurement of additional staff. • Acquisition of new commercial clients in the transport sector