SNCF, the french railway company, has been using OpenStreetMap to map railway stations for years. Since 2016, travelers can explore all 380 railway stations of the Greater Paris area through interactive maps deployed on their web site and on a mobile app.

Browsing a map is good, but not good enough when you are lost inside a complex, multilevel, indoor (and/or outdoor) environment. It gets worse for someone in a wheelchair, elderly people, or anyone coming back from a family trip with more luggage than your hands could carry.

What if we could bring pedestrian routing and navigation services to help travelers find their way inside and around railway stations? Could we do it using OSM data?

This is the challenge SNCF has been trying to tackle for the past few months. Together with Jawg Maps and Carto’Cité, they held a detailed mapping campaign on 83 stations and developed a dedicated routing engine. This engine can seamlessly navigate indoors and outdoors, and achieves routing through open spaces such as pedestrian areas and station halls.

This talk exposes the challenges we faced. It focuses on the data structure and the algorithmic strategies that have been defined to provide a robust navigation service, and its ability to handle the great diversity of railway stations. The talk also discusses how some parts of the data model originally designed for road navigation could be used for pedestrian routing – with some minor tweaks.

If you thought mapping stairs was trivial, you would be surprised…