Challenges and experiences in using heterogeneous, geo-referenced data for automatic creation of driving simulator environments

摘要

For the development of advanced driving assistance and automation systems the simulation plays an important role. Urban areas get increasing emphasis, especially in the context of future Car2X-communication. This article describes an approach developed in the project Virtual World. Its goal is to model such virtual three-dimensional (3D) environments and logical road descriptions automatically based on a tool chain from heterogeneous geographic datasets (e.g., cadastral data, road surveying, aerial pictures, and crowd-sourced data). As proof of concept, the urban area of Braunschweig, Germany, was chosen. The article focuses on the generation of a 3D city model with corresponding road network description suitable for driving and traffic simulations as used in research and industry. The article gives technical descriptions of the major work steps and discusses issues regarding the availability of data. It concludes with the current project outcome and further development in the project.