The latest Swiss high-tech was showcased at this year’s Mass Trans Innovation Japan as the cooperation of the Swiss Federal Railways (SBB) and Swiss Center for Electronics and Microtechnology (CSEM) presented their newest AI application in track maintenance. During a seminar at Mass Trans Innovation held on November 29, Marcel Zurkirchen, Head of Measurements and Diagnostics at SBB, and Philipp Schmid, Head of Robotics and Machine learning at CSEM, elaborated on the collaboration to develop an intelligent detection system that is to revolutionize the process of inspection in the railway sector.
SBB is the national railway company in charge of operating the Swiss railways track system and its maintenance. Bearing the responsibility for the supervision of over 3’000 km of tracks, SBB started to develop a system to locate defects on the tracks by using inspection wagons. These wagons collect huge amount of data by taking pictures of the tracks condition with ultra-speed cameras and add supplementary information such as weather conditions and GPS data. To process those vast data sets, SBB turned to CSEM for cooperation.
CSEM is a Swiss research and development center that bridges industries and academia to provide technical solutions in the areas of micro technology, nanotechnology, microelectronics, systems engineering, photovoltaics and communications. One of the focuses of its local branch in Alpnach lies in AI, particularly in deep learning. By mimicking the neuronal layer structure of the human brain, large data sets can be processed to recognize patterns, thus creating intelligent systems for image recognition for instance.
In the case of the images generated by the inspection wagons of SBB, CSEM trained their software to locate defects on the tracks and differentiate between a real defect and what would for example be a chewing gum stuck on a track. The AI can discern between various categories of defects –such as mistakes in wielding or damages caused by erosion- and by overlaying data collected at later points in time, the software can track the development of a particular defect.
The detection AI was presented in the context of a seminar organized by the Swiss Business Hub Japan during the Mass Trans Innovation Japan 2019. Mass Trans Innovation Japan is the only comprehensive trade show specialized in railways technology in Japan and took place from November 27 until 29, 2019.