Peter Vortisch, Karlsruhe Institute of Technology (Germany)
Simulation of Traffic and Mobility
Peter Vortisch is Professor of Transportation at the Karlsruhe Institute of Technology and the head of KIT’s Institute for Transport Studies. His main research area is modelling travel demand and traffic flow.
ABSTRACT:
In the field of mobility, seen from the point of view of Civil Engineering, computer simulation is used in two main areas: travel demand forecasting, i.e. estimating how much traffic we have to expect on our transportation infrastructure, and capacity analysis, i.e. finding out if the given infrastructure will be able to handle the expected demand. For both areas, a variety of models is used in transport planning and traffic engineering, and in both areas the models face new challenges.
In transport planning, typically the traffic during one day of a whole city is modelled using macroscopic travel demand models. In research-oriented environments, microscopic, i.e. agent-based simulation models are used, representing each traveling person in the area individually. These models provide a better basis to include innovative mobility services such as car sharing, ride hailing etc. One of the aspects currently under research is the stability of the travel behavior during a week, i.e. how the individual choices of destinations or transport modes at the single days depend on each other. Empirical data show that a week is more than just seven days, and there are different approaches how to implement that in the models.
The traffic flow simulation models used in traffic engineering have a long tradition going back to the 1970ies, but there are still a lot of challenges to meet. One is to reproduce complex driving situations such as merging maneuvers in congested traffic including cooperation of drivers. Another rather new aspect is that traffic flow simulation will be more and more used to test the control software of automated vehicles. Traditionally, in traffic engineering the road capacity was the most important outcome of the flow simulation, and to reproduce capacity exactly, tradeoffs were accepted such as unrealistic acceleration values. For testing automated vehicles or driver assistant systems, capacity is not of importance, but realistic individual behavior is. Most available simulation tools still have to adapt to this shift in requirements.