Last weekend I attended an OpenStreetMap workshop about indoor mapping hosted by the German Federal Agency for Cartography and Geodesy in Frankfurt. After having started to use OSM indoor data in 2020 for KDE Itinerary this was my first opportunity to actually meet other people working on that subject in person.

OSM Indoor Data in KDE Itinerary

Indoor data here refers to mapping information about the inside of a building, which compared to outdoor map data tends to have an even higher need on precision (centimeter-level rather than meter-level), and much more often involves a third dimension (the floor level). In KDE Itinerary we use this to display detailed maps for train stations and airports.

Screenshot of KDE Itinerary's train station map showing parts of Frankfurt central station.
Parts of Frankfurt central station displayed in KDE Itineary.

The OSM data model for this has been designed under the assumption of having a single self-contained building. Large train stations can however cover several city blocks, overstretching this model quite bit. That’s one of the topics I hoped to get input on during the workshop, and it turns out we are indeed not the only ones running into that problem.

Another issue we encountered is modeling of underground structures. While many elements (say a train track) can exist on their own above ground, they only really make sense underground with a corresponding “cut-out” around them (say a tunnel).

Different use-cases, very similar challenges

Displaying train station maps is just one possible thing to do with OSM indoor data, many other use-cases were represented at the workshop as well.

Photo of attendees of the OSM indoor data workshop.

While that is a very diverse set of perspectives, a lot of this is relevant for features we sooner or later want in KDE Itinerary as well. Positioning, routing and 2.5D or 3D views are on the wishlist for our train station maps, and accessibility is something we want to focus on more as KDE as a whole as well.

And in any case, pretty much everyone is facing the same challenges regarding data modelling, data quality, data level of detail, processing algorithms, you name it. It’s certainly extremely valuable to have a forum to exchange on those topics.


One and half a day workshop wasn’t even remotely enough to cover all topics, so there are now online meetings planned to continue the discussion, as well as another in-person meeting at the FOSSGISS conference in March in Berlin.