SBB is currently checking the reliability of rail network maintenance using artificial intelligence (AI). According to a press release, SBB has entered into a partnership with the Swiss Federal Institute of Technology in Zurich (ETH), the ETH spin-off LatticeFlow, and Siemens to do so.
SBB operates one of the densest railway networks in the world, with 7,500 kilometers of tracks and 300 tunnels serving over 1.25 million travelers daily. Constant maintenance and testing of this network had until now been conducted by employees and on-site visual inspection. In future, artificial intelligence is to be utilized for this purpose, for instance by driving special vehicles with cameras to visually inspect the tracks. An assessment can then be made with AI. Since this concerns the safety of the network, it must be ensured that the AI models provide reliable results not just under laboratory conditions but also resiliently in practice. Scratches on the lens, raindrops or snow on the tracks and other variables that impact the image quality must not affect the reliability of the AI system.
Ilir Fetai and Andre Roger, who lead the Center of Competence in Machine Perception at SBB, are ensuring that these plans are achieved and that the SBB’s position as a leader of innovation in the railway sector is strengthened. Fetai is quoted as follows: “Artificial Intelligence (AI) is one of the core subjects of SBB as we see great potential with its use for improved, smart, automated monitoring of our railway infrastructure. Our collaboration with LatticeFlow, the ETH and Siemens in the area of robust, reliable AI is a decisive factor in our ability to fully exploit the use of AI.”
The SBB’s AI models will first be assessed using the reliable AI platform from LatticeFlow, a company based in Zurich, and they will be improved if necessary. The aim of the collaboration is to prove that the AI models are reliable and can be safely implemented in practice.
LatticeFlow was formed as a spin-off from the ETH in 2020. The start-up supports companies in developing reliable solutions in the area of artificial intelligence.