Instead of sending staff to walk the tracks and inspect them by eye, Swiss Federal Railways (SBB) uses specially equipped diagnostic trains and artificial intelligence to examine around 2,000km of track each month, according to SRF.
Rather than inspecting the railway at walking pace, the trains travel at up to 120kph, measuring and filming the track as they go. AI then analyses the images to identify damage and defects.
The technology is not new. SBB began experimenting with AI for track inspection about a decade ago, but the early results were disappointing: the system produced too many errors. Advances in AI have since made it more accurate and reliable.
SBB now operates two diagnostic trains equipped with the technology. Together, they inspect about 2,000km of track a month and detect as many as 1,500 defects, most of them minor.
The trains carry a range of instruments. Track-geometry measuring devices, for example, can detect minute deviations in the rails, while cameras record images that AI systems can analyse for visible faults.
Yet AI will not make human inspectors redundant. The system is trained to recognise particular defects in specific parts of the track. Problems outside those areas—for example, damage to an overhead-line mast—will not be detected.
Trained human eyes are therefore still needed to spot what the machines are not aimed at. And trained hands are needed for repair work.
More on this:
SRF article (in German)
For more stories like this on Switzerland follow us on Facebook and X.
