The KAUST Scene Generator is a software that can generate three-dimensional road networks from freely accessible OpenStreetMap data. This means that freely usable but difficult to process geodata will be available in the future as a source of content for driving simulators or as data sets to train self-driving cars.
3D street model of Mariahilfer Straße.
The algorithm for the generation of intersections produces geometrically meaningful results even from complex original data. Where the corresponding data is available, road markings are also generated.
The KAUST Scene Generator can be used as a plugin of the OpenStreetMap Editor JOSM.
The street network for entire city districts can be created with just a few clicks. If an elevation model and aerial photographs are available, they are integrated.
On behalf of the Saudi Arabian King Abdullah University of Science and Technology (KAUST), VRVis has developed software that can generate 3D scenes from open-source geodata. The project's goal was to find ways to generate affordable or even free content for the existing KAUST driving simulator. What is special about the prototype of the KAUST Scene Generator written in Python is that, for the first time, data from the free geoinformatics platform OpenStreetMap is processed in such a way that comprehensive 3D generation is possible. Furthermore, data from height maps as well as satellite or aerial images can be integrated into the calculation, which enables an even more realistic 3D model. The most innovative feature of the KAUST Scene Generator is the creation of roads and intersections: OpenStreetMap uses a graph display in which streets are displayed as sequences of points with attributes and in which, as a rule, no specific street widths are available. Despite challenges like these, the KAUST Scene Generator automatically calculates curve roundings and geometrically logical intersection solutions.
The 3D models of the KAUST Scene Generator are optimized for Unreal Engine 4, in which the already existing KAUST driving simulator was programmed, but can also be easily exported to a number of other formats, which opens up further fields of application in addition to their use in various simulation settings. For example, computer-generated road networks could serve in the future as cost-effective alternatives or additions to the training of self-driving cars or as scenarios for personalized computer game worlds.