In today’s world of digitalisation and Artificial Intelligence, there isn’t a single industry that isn’t concerned by these technologies. Given that R&D Vision operates in various areas to design dedicated optical measurement systems, the company is regularly contacted to design digital maintenance assistance systems for non-destructive testing (NDT), 3D terahertz reconstruction, infrared thermography and so on.
Eurotunnel adheres to this digitalisation dynamic and has been working for more than three years with R&D Vision to design customised image acquisition and processing systems that meet Eurotunnel’s trade requirements.
Eurotunnel and R&D Vision recently designed a measurement van specifically for analysing the catenary and civil engineering (roof). This van assists the maintenance teams in performing fine diagnostics and anticipating certain malfunctions. It has been deployed into commercial service on the trains used to transport lorries that travel at 140 km/h. As a result, a precise scan is taken of the 110 kilometres of the Eurotunnel.
To date, this truck has specific lighting designed by R&D Vision and three 4K high-speed colour cameras with a data transfer rate of more than 2 GB/s and a resolution of 0.5 mm/pixel around the objects observed. The design and manufacturing as a whole by R&D Vision (lighting, cameras, raw data monitoring and acquisition IT rack and software) is scalable both in terms of hardware and software. It is therefore entirely realistic to consider integrating other colour cameras to have 3D visualisation or even to use other types of cameras to detect hot spots and wet areas.
When it comes to image processing, this is currently performed by operators. It’s an arduous task to analyse the 7 TB of data generated on a return trip. The image quality in terms of resolution, colour rendering and camera angle reproducibility means that different types of algorithms (based on contour analysis, 2D-3D matching and, of course, Artificial Intelligence with deep learning) can be deployed to automatically detect defects between two runs.
An intuitive Human Machine Interface that groups trade requirements was developed for data acquisition and review to ensure straightforward, robust deployment. Given that nobody is allowed to remain in the truck while the measurements are taken, the system is remotely controlled using a 4G or WiFi connection.
With entry into service at the beginning of 2019, the truck will be used once a week by the Eurotunnel teams. Bearing in mind that 7 TB of data is collected a week, significant storage capacities will be required to ensure proper traceability!