Agents for Traffic Simulation

Vehicular traffic is a classical example of a multi-agent system in which autonomous drivers operate in a shared environment. The article provides an overview of the state-of-the-art in microscopic traffic modeling and the implications for simulation techniques. We focus on the short-time dynamics of car-following models which describe continuous feedback control tasks (acceleration and braking) and models for discrete-choice tasks as a response to the surrounding traffic. The driving style of an agent is characterized by model parameters such as reaction time, desired speed, desired time gap, anticipation etc. In addition, internal state variables corresponding to the agent's "mind" are used to incorporate the driving experiences. We introduce a time-dependency of some parameters to describe the frustration of drivers being in a traffic jam for a while. Furthermore, the driver's behavior is externally influenced by the neighboring vehicles and also by environmental input such as limited motorization and braking power, visibility conditions and road traffic regulations. A general approach for dealing with discrete decision problems in the context of vehicular traffic is introduced and applied to mandatory and discretionary lane changes. Furthermore, we consider the decision process whether to brake or not when approaching a traffic light turning from green to amber. Another aspect of vehicular traffic is related to the heterogeneity of drivers. We discuss a hybrid system of coupled vehicle and information flow which can be used for developing and testing applications of upcoming inter-vehicle communication techniques.