The emergence of advanced driver information systems sets new challenges in the area of traffic modelling and simulation. The use of techniques such as autonomous route guidance, advanced incident warning or information channels dedicated to traffic information must be incorporated in a new generation of traffic models. This paper first discusses the impact of these new systems on modelling methodology, with an emphasis on some important issues to make the new models both realistic and adaptable to technological and sociological evolution. The main issues are the avoidance of the "steady state", "perfect information" and "uniform behaviour" assumptions commonly made in traffic simulation. The need of an explicit model for traffic information flow and of behavioural routing methods allowing for the real time use of this information are introduced as consequences of this first conclusion. These in turn have consequences on more technical levels, such as network representation and traffic metaphors. The argumentation developed is then illustrated using the definition of some of the modelling concept within the PACSIM model for traffic assignment. The dynamic nature of the simulation, the concepts of "packets" (considered as a traffic metaphor) and multi-level network, the structure of the information network (as distinct from the road network itself) and finally the bi-level behavioural routing methodology applied in PACSIM are shown to be coherent with the requirements discussed in the first part of the paper.
Abstract