Modeling aggressive driver behavior at unsignalized intersections

Abstract

The processing of vehicles at unsignalized intersections is a complex and highly interactive process, whereby each driver makes individual decisions about when, where, and how to complete the required maneuver, subject to his perceptions of distances, velocities, and own car's performance. Typically, the performance of priority-unsignalized intersections has been modeled with probabilistic approaches that consider the distribution of gaps in the major-traffic stream and their acceptance by the drivers of minor street vehicles based on the driver's “critical gap”. This paper investigates the aggressive behavior of minor street vehicles at intersections that are priority-unsignalized but operate with little respect of control measures. The objective is to formulate a behavioral model that predicts the probability that a driver performs an aggressive maneuver as a function of a set of driver and traffic attributes. Parameters that were tested and modeled include driver characteristics (gender and age), car characteristics (performance and model year), and traffic attributes (number of rejected gaps, total waiting time at head of queue, and major-traffic speed). Binary probit models are developed and tested, based on a collected data set from an unsignalized intersection in the city of Beirut, to determine which of the studied variables are statistically significant in determining the aggressiveness of a specific driver. Primary conclusions reveal that age, car performance, and average speed on the major road are the major determinants of aggressive behavior. Another striking conclusion is that the total waiting time of the driver while waiting for an acceptable gap is of little significance in incurring the “forcing” behavior. The obtained model is incorporated in a simple simulation framework that reflects driver behavior and traffic stream interactions in estimating delay and conflict measures at unsignalized intersections. The simulation results were then compared against real world observations, representing an important step in validating the model of aggressive driver behavior at unsignalized intersections.