MODELING DRIVER DECISION MAKING: A REVIEW OF METHODOLOGICAL ALTERNATIVES
Fred Mannering University of Washington
Research in the general area of Intelligent Transportation Systems (ITS) has given rise to important questions regarding drivers' decision making, which has always been fundamental to understanding vehicle traffic flow and to developing models for predicting urban traffic congestion. However, because of the complexities involved in modeling and understanding drivers' decision making, standard urban transportation modeling has often focused on a more aggregate level: viewing traffic flow on highways as an observational unit to be studied. This has led to an entire body of literature that considers traffic flow to be roughly analogous to fluid flow. This literature applies principles of fluid flow, such as shock-wave analysis ( May, 1990), to the modeling of traffic flow. Such an approach is an attempt to replicate the product of individual driver decision making and has been useful in many applications. However, in the presence of rapidly changing technology, such as that offered by ITS, the focus of research must be directed toward the primary decision-making unit--the driver--because the standard fluid-flow analogy is not likely to apply in an environment of vehicles containing possibly different levels of this technology in a single traffic stream.
To be sure, there has been a considerable amount of research on driver decision making and its ultimate impact on traffic flow. This research has been conducted in a variety of disciplines including human factors, transportation engineering, and accident analysis. However, models of driver decision making have yet to be integrated successfully into a more general