Traffic Simulation

The actual traffic situation clearly influences the planning of the driving strategy and the current driving behaviour. Accordingly, it is important to consider traffic as an external environment or boundary condition in the simulation-based prediction and analysis of vehicle behavior. This is achieved by adequate dynamic traffic models in the simulation process.

There are three main approaches to traffic simulation:

  • macroscopic traffic models
  • mesoscopic traffic models
  • microscopic traffic models

These are each extended with stochastic modelling techniques in order to map variabilities. Depending on the application, we use all common modelling approaches and are also particularly concerned with the further development, improvement and combination of different approaches.

Our software VMC® uses a stochastic macroscopic traffic model developed and implemented by us. This model allows to assess and to quantify the influence of traffic, for example on drive train loads, energy requirements and consumption, and also to systematically investigate sensitivities and dependencies.

The parameterization of traffic models generally represents an important challenge and is therefore another focus of work at our institute in this area. For the provision and parameterization of traffic models in VMC® several options and supporting techniques are available to the user. An optional interface to the here® database is particularly noteworthy, which makes it possible to retrieve historically recorded average speeds on any available route depending on the time of day and week. This procedure then also allows, for example, time-of-day-dependent analyses with regard to energy demand and consumption.

Verkehrsplanung Kreuzung
© Fraunhofer ITWM
Verkehrsszenario in einer Kreuzung

Microscopic Traffic Simulation for Testing ADAS/AD

In the field of traffic simulation, we also provide dynamic traffic scenarios for testing and developing driver assistance systems and ADAS/AD functions. In order to test and to evaluate such assistance systems systematically and efficiently, it is essential to be able to virtually simulate realistic driving scenarios. This includes in particular complex traffic situations as well as a realistic, statistically founded analysis of possible variations and variabilities. We use microscopic traffic models embedded in realistic scenes such as road networks, intersections or roundabouts and combine them with stochastic techniques to reflect the variabilities.

The models and scenarios can be used in purely virtual studies or coupled with offline simulations or with interactive driving simulators. Here we offer with our own driving simulator RODOS® the possibility to investigate such analyses and studies with human drivers in different mixed traffic scenarios.