Simulations of highway chaos using fuzzy logic

We report on simulations of chaotic traffic flow in freeways based on a fuzzy knowledge based model of driver behavior. Our simulation approach is a microscopic method similar to Q. Yang and H.N. Koutsopoulos (1996), where individual vehicles are separately modeled. The knowledge base consists of an extensive and exhaustive set of fuzzy IF-THEN rules depicting all possible traffic situations that a driver is likely to encounter in a highway. The principal motivation behind using a fuzzy knowledge based approach to model the driver behavior is because fuzzy modeling provides an effective means to break up any highly nonlinear system into IF-THEN rules (J.M. Mendel, 1995). In addition, fuzzy logic is well equipped to handle uncertainties that are present in real world traffic situations (P. Chakraborty and S. Kikuchi, 1992; S. Kikuchi and M. Pursula, 1998). Therefore using fuzzy rules, it is possible to incorporate linguistic descriptions of scenarios such as 'speed is moderate' or 'adjacent lane gap is quite acceptable'. Instead of quantifying variables into crisp classes as in a traditional expert system, this is the manner in which a driver is more likely to perceive any situation. Fuzzy logic techniques are receiving recent attention as important tools in transportation engineering studies and such approaches have met with a great deal of success.