While the focus of semantic 3D city models has so far been on 3D building models, the increasing availability of detailed representations of the street space allows numerous new applications. The international OGC standard CityGML which is used to model and exchange semantic 3D city models has been improved to cover street spaces. Version 3.0 includes extended concepts for modelling the street space. This thesis investigates how CityGML 3.0 compliant street space models can be used for multimodal navigation applications by mapping them to a graph database. In particular, the concepts required for multimodal navigation are being analyzed. Further, it is analyzed if those apply to CityGML 3.0. The combination of different means of transport in one application and the use of different dimensions in the geometric representation as well as their "level of detail"/granularity also play an essential part here. Firstly, requirements for multimodal navigation applications are collected. Then, concepts of CityGML 3.0 are compared with these requirements. Missing elements, such as adjacency relationships and weights can be added to a graph representation. This is followed by the pre-processing of the CityGML test data in the graph database Neo4j. Using the graph database, a suitable network for multimodal routing is generated. The final routing network shall connect existing CityGML objects without removing elements or canceling existing relationships. In the following, different weights are added to the newly generated network, for example, length derived from the geometry or speed limits derived from semantic information of CityFurniture objects (street signs). The resulting multimodal network is then used to show that the analysis of shortest paths is independent of the geometric representation and the granularity of the street space objects. For the multimodal navigation application, two CityGML 3.0 test datasets are available (Ingolstadt and Grafing near Munich). To verify the 3D navigation capabilities, a new dataset containing the Grafing network as well as a hand-modelled 3D parking garage building with differing granularity is used. Finally, the routing results are validated. Street space data from CityGML 3.0 can provide rich information for navigation applications. In addition to the geometric and semantic information, further connections between elements can be used. Furthermore, CityGML allows the modelling and use of different representations of the street space, for example, using the complete road or a lane-based representation. A network structure based on the predecessor and successor relationships of the TrafficSpace objects can be generated independent of the remaining structure and scope of the dataset. Furthermore, the network structure can be improved with additional connections. For querying the data and performing shortest path analyses, the graph database is an efficient platform. Fundamental multimodal navigation functionality could be implemented. However, there is a need for further development, e.g., the integration of additional transportation modes. In a real-world example, the combined usage of different granularities and dimensions of TrafficSpace objects was shown. Vehicle routing from the street to a parking space in the multi-story car park was carried out. It is then possible to continue the route from the parking spot to the footpath network. Findings from this work have also contributed to the further development of the open-source software r:trån.     «

While the focus of semantic 3D city models has so far been on 3D building models, the increasing availability of detailed representations of the street space allows numerous new applications. The international OGC standard CityGML which is used to model and exchange semantic 3D city models has been improved to cover street spaces. Version 3.0 includes extended concepts for modelling the street space. This thesis investigates how CityGML 3.0 compliant street space models can be used for multimoda...     »