Over the past few years I've performed research in the simulation of traffic, both vehicular and pedestrian. In both cases, however, simulations are of the so-called microscale type. This means that the system as a whole was not directly simulated but only its constituent components. The vehicular simulations are much simpler and for them each vehicle follows a small set of strict rules making up a cellular automaton.
In the case of pedestrian traffic, multiple types of agents are instantiated, for instance, adults, children, stores, information signs, etc. At the most complex stage, animated agents are able to learn from the environment and make decisions based on acquired data. Decision making and movement is a layered process, which is described in detail in subsequent web pages. No need to bore you with it now!
In summary, the vehicle simulations are interesting because of the complex emergent behavior resulting from a very simple rule set. The pedestrian simulations are far more interesting because of the greater complexity required to navigate a pedestrian environment.