UBC Theses and Dissertations
Modelling the driver for traffic flow simulation Stang, Norman Erik
For 60 years, engineers have modelled traffic flow for use in roadway analysis and design. A continuing problem with such models, though, is their inability to adequately capture the human element in the system. The human driver does not think and act in precise ways, making him difficult to model using conventional mathematical means. This research explored an alternative method of formulating driving models. It started from a psychological basis and used fuzzy logic. Fuzzy logic provided a systematic way of handling imprecision, and its constructs allowed a more intuitive model of driving, one more closely resembling the thinking and acting patterns of humans. A new fuzzy logic driver model was developed. Its structure was based on a general psychological model of human information processing. It was designed so it could be programmed with specific driving behaviour by an end user. To demonstrate and validate it, the model was programmed for two lane rural highway driving and used in simulations of these facilities. Results were compared to field data and the Highway Capacity Manual and proved favourable. The result of the work was a computer based road simulation toolbox containing the new fuzzy logic driver model. The toolbox provides a user with the ability to construct his own road networks, driver types, and vehicle types. With these, he can simulate traffic and examine both isolated incidents and overall performance measures.
Item Citations and Data