Correct positioning of rail clips using AI camera

Mon 16 May 2022 17:32

In 2019, Pandrol participated with a challenge on Business Innovation Day. The research project EDGY is now starting, where Mid Sweden University and Pandrol will develop a smart camera for track-laying machines.

En bit räls i en pallkrage, omgiven av sten och ett clips är synligt. Ovanför syns en kamera som är riktad mot clipset.

In 2019, Pandrol was at Business Innovation Day (BID). They participated with a challenge of quality assurance of railway rail laying where researchers from STC participated as experts.

Pandrol came into contact with Business Innovation Day via Christoffer Andresen at RISE/Propell in Hudiksvall.

– They contacted us and asked if we had an idea that we wanted to discuss with experts, and that’s the way it is, says Anders Olsson, Global Equipment Electrification Manager, at Pandrol.

The challenge they worked on was whether the detection of obstacles around a railroad was possible with the help of object recognition.

In 2019 when the discussion first started.

Experience from participating in BID

– Participating in Business Innovation Day has given us a chance to participate in a research project based on the original challenge. We were supposed to write an application back in 2019, but since we have foreign owners, possible financiers were limited, which was a challenge. So, it has taken some time to move from idea to actual project, explains Anders.

On March 1, the research project EDGY started, where Pandrol participates together with researchers Irida Shallari and Mattias O'Nils from Mid Sweden University. What started as an idea during Business Innovation Day 2019 is now an ongoing project at Mid Sweden University.

The project will develop a smart camera, with the help of AI and deep learning, which will be used to ensure the quality of rail laying. The smart camera is meant to be mounted on track-laying machines to get real-time images of, for example, mounting the clips that attach the rail.

What are Pandrol’s goals and hopes with the project?

– Our goal is for the project to result in new competitive services and products for us and that we get a chance to learn what is possible with object recognition. This will make it easier for those who use our machines, Anders concludes.


The page was updated 5/18/2022