MODELING BEHAVIOR IN VEHICULAR AND PEDESTRIAN TRAFFIC FLOW by

This dissertation investigates the design and analysis of vehicular and pedestrian models. A type of vehicular model is developed both to offer novel contributions to vehicle behavior modeling as well as to use as a tool to learn how to create an even more complex behavioral model of pedestrian movement. First, the current state of modeling is investigated including purely behavioral studies and engineering modeling techniques. Behavioral studies are drawn largely from the field of urban affairs and planning while engineering modeling methods are drawn from civil engineering, mathematics, and computer science. Second, a model of vehicular traffic is constructed by first implementing existing work in software. Existing work focuses on single lane traffic, so we next extend the model to support lane changing in multiple lanes. The new mathematical rules are implemented in software and effects of lane changing then studied. The model contributes new capabilities to the field and provides experience to next create a more complex pedestrian model. Third, an algorithmic model of pedestrian movement is created. At its simplest level, steering rules are used that are drawn from the literature. New rules and models are created to support groups and simple social interaction. Learning and memory are then modeled so that simulated pedestrians are human-like in ways that have an effect on congestion. Fourth, software is developed that implements the model. While the model offers a means, i.e., function parameters, for calibration, an implementation must exist to take advantage of that. The software is designed using an object-oriented