For details, see the paper in the proceedings: https://doi.org/10.5281/zenodo.21342252

The geodata in OpenStreetMap (OSM) is mainly collected and used for the purpose of providing orientation and guidance in public outdoor spaces in both urban and rural areas. To this end, dedicated geospatial applications such as cartographic maps and routing applications consume the OSM data. Many indoor spaces within buildings form closed islands within the public space and are not easily accessible and thus not interesting for OSM. In other places however, public space extends into buildings either through open entries or because spaces are only partially enclosed by built structures, forming a gradual passage from outdoor to indoor with half-open areas. Recently, maps and routers are increasingly extended to also cover publicly accessible indoor spaces, similarly to outdoor spaces. Subsequently, the data gathered needs to extend to the indoor and include the interior structure of buildings, rooms and walkable areas, paths through corridors and staircases. There has been a niche interest in indoor mapping for a long time already. Respective work dates back to the early 2010s, with proposals for tagging schemas from the academic field (Goetz and Zipf 2011) as well as the mapping community (Poole et al. 2014). Recently, there is also a growing interest to use data from architectural design, construction processes and real estate management as a base of indoor mapping (e.g. Krishnakumar, Tauscher, and Heigener 2023).

While outdoor and indoor spaces are not fundamentally different, and hence mapping and data are similar, there are notable differences. Indoor spaces have a more detailed structure compared to natural environments or infrastructure features, but similar to urban outdoor spaces designed for humans to stay in. Further, buildings, containing indoor spaces, consist of stacked walkable areas in multiple levels - similar as grade-separated street and rail infrastructure but employed to a greater extent with usually more than two levels on top of each other. All indoor relations, areas, ways as well as nodes for POIs are attached to these levels.

Currently, the indoor data is sparse in OSM with usually only few prominent buildings in a city mapped through dedicated mapping campaigns or coordinated efforts in research projects. According to taginfo.openstreetmap.org, there are over 700~million objects tagged as buildings, but only 1~million indoor objects and a typical building contains various indoor objects. However, to unlock the opportunities and potentials of new applications for integrated indoor-outdoor visualization, navigation and analysis, we would need an extensive survey and subsequent insertion of data about all kinds of indoor spaces into the OSM database. Given the more detailed nature of this data with intricate representation of interior structures and the fact of stacked areas multiplying the amount of data within a given footprint area, we assume that the increase of data in the database would be substantial and exert a notable effect on the OSM infrastructure, the underlying databases and network traffic. To judge these effects, it would be interesting to know which volume of indoor data would be created if all possible indoor spaces were comprehensively mapped in OSM. There is - to the best of our knowledge - no study yet that tries to quantify the potential amount of indoor data to be mapped in OSM. With this study we aim to estimate this potential and as a side result we can make a statement about the completeness of indoor data, similar to earlier studies that evaluate quality and completeness of building data (Biljecki, Chow, and Lee 2023).

We operate under the premise that only publicly accessible spaces are relevant for inclusion in the OSM database. At the same time, only very few buildings are comprised entirely of public spaces. Thus we must not only identify buildings that have a public function, but also estimate the portion of spaces within those buildings that are actually publicly accessible. We base our estimation of "mappable" indoor spaces on an areal (square) measure, hence footprint areas of buildings, levels and rooms or open spaces in buildings. First, we identify various classes of buildings that are publicly accessible, at least partially, such as stations, malls, universities, hospitals, town halls etc. Second, we apply a heuristic method to estimate the percentage of publicly accessible areas for these classes of buildings. In a third step, we extract the relevant building data for a limited geographical scope, calculate the total building area within that scope across all buildings with all their levels. Finally, by applying the percentages from the second step, we calculate the estimated indoor mapping potential. We can then compare this value with the actual indoor data in OSM aggregated across the chosen geographical scope or building by building to retrieve some key figure of completeness.

... full paper in proceedings: https://doi.org/10.5281/zenodo.21342252

Biljecki, Filip, Yoong Shin Chow, and Kay Lee. 2023. "Quality of Crowdsourced Geospatial Building Information: A Global Assessment of OpenStreetMap Attributes." Building and Environment 237: 110295. https://doi.org/10.1016/j.buildenv.2023.110295.

Goetz, Markus, and Alexander Zipf. 2011. "Extending OpenStreetMap to Indoor Environments: Bringing Volunteered Geographic Information to the Next Level." In Urban and Regional Data Management, edited by Sisi Zlatanova, Hugo Ledoux, and Elfriede Fendel, 51–62. https://doi.org/10.1201/b11647-7.

Krishnakumar, Subhashini, Helga Tauscher, and Dominik Heigener. 2023. "Floor Plan Extraction from Digital Building Models." In Proceedings of FOSSGIS (Academic Track) 2023, 146–52. Berlin, Germany. https://doi.org/10.5281/zenodo.7576205.

Poole, Simon, Tobias Knerr, Peda, and Andreas Hubel. 2014. "Simple Indoor Tagging." https://wiki.openstreetmap.org/wiki/Simple_Indoor_Tagging.