Traffic Flow Modeling Using Available Cloud-Based Traffic Velocity Information

Due to a constantly increasing interest in autonomous driving, the development of advanced driver assistance systems plays an important role in automotive industry. Current research activities try to include information on the current traffic situation, for example, to improve predictions or assistant functions and emphasize the need for more accurate traffic data. In this sector cloud services already offer basic information about the traffic state such as the current traffic speed. Since online traffic data is partly incomplete or availability cannot be guaranteed for every position, model-based approaches present an option to overcome this problem. Traffic flow models are a useful tool to reproduce traffic situations given the necessary initial and boundary conditions. Within the scope of this work, online traffic data is used to predict the mean traffic velocity for an upcoming road segment by means of a second-order partial differential traffic model. Using highway detector measurements for validation, it is shown that simulations provide meaningful results for traffic speed calculation. Thus, it is possible to predict traffic states spatially for an upcoming horizon. The benefit of this study is reflected in the ability to simulate traffic scenarios not only offline but also online using cloud-based data and consequently making it deployable for in-vehicle applications.