A Crowdsourced Indoor Modeling for Smart Cities: Tackling Complex Building Geometries with IndoorGML

This study presents an automated approach for generating IndoorGML-compliant navigation graphs in complex indoor environments using volunteered geographic data from OpenStreetMap. The method addresses challenges in modeling non-convex and irregular building geometries by applying constrained triangulation, followed by merging adjacent triangles into convex hulls. These convex regions simplify the geometry, reduce redundant nodes, and improve the coherence of the navigation graph. Indoor spaces are represented using the IndoorGML core model, where centroids of convex regions serve as graph nodes connected via doors using Euclidean distances. The approach is tailored to overcome geometric noise and tagging inconsistencies in VGI datasets. A case study on a university building demonstrates the method's effectiveness in producing accurate, compact, and standards-based navigation graphs. The use of open access OSM data further reduces mapping costs and supports scalable modeling for applications in indoor routing, emergency management, and smart building systems.

کلیدواژه ها:

Indoor Navigation ، Smart City ، complex building ، IndoorGML