In recent years, digital twins and semantic 3D city models have become increasingly important for urban analysis, simulation, and planning. Among the available standards, CityGML 3.0 offers a detailed and flexible framework for representing buildings, including their interior spaces. Despite this, generating semantically structured indoor models in an automated way remains difficult, especially when working with heterogeneous data sources such as georeferenced point clouds and 2D CAD floor plans. Many existing methods rely either on concentrating mainly on geometric reconstruction or heavily on manual editing, which limits their scalability and practical usability. In order to create and enhance CityGML 3.0 indoor models, this thesis suggests an automated workflow that combines point cloud data with CAD-based floor layouts. The fundamental idea is to combine height information taken from 3D point cloud data with reliable boundary information from 2D floor layouts. This method allows for the more constant reconstruction of structural elements such walls, floors, ceilings, and openings. Coordinate alignment, vertical dimension derivation, spatial relationship detection, and the creation of hierarchical and topological structures necessary for the CityGML indoor model are all addressed by the created processing pipeline. The workflow focuses primarily on generating the building’s outer shell at Level of Detail 2 (LoD2) and indoor building blocks at Level of Detail 1 (LoD1). Also, exploring how this approach could be extended toward more detailed room-level representations at LoD1. Automated feature extraction, semantic enrichment, and the correct representation of spatial relationships among building elements are key to this approach. The overall goal of the suggested method is to facilitate the creation of indoor models for digital twin applications that are more scalable and repeatable. The findings show that a workable method for creating semantically structured, standards-compliant CityGML indoor models while greatly lowering the requirement for manual modeling procedures is to integrate 2D CAD geometry with 3D point cloud data.     «

In recent years, digital twins and semantic 3D city models have become increasingly important for urban analysis, simulation, and planning. Among the available standards, CityGML 3.0 offers a detailed and flexible framework for representing buildings, including their interior spaces. Despite this, generating semantically structured indoor models in an automated way remains difficult, especially when working with heterogeneous data sources such as georeferenced point clouds and 2D CAD floor...     »