Modeling and Empirical Analysis of Tailgating Behavior of Drivers
Shrestha, Deepak Kumar
Lovell, David J
MetadataShow full item record
This dissertation presents a microscopic study of tailgating behavior of drivers. There are very few studies focused on tailgating, although it is a serious issue for traffic safety. The reason for very few studies might be the fact that tailgating is a complex problem involving human behavior and kinematics of the vehicle and it is also equally challenging to collect naturalistic driving data relevant to tailgating. Because this approach is empirical, we developed a sophisticated data acquisition system using an instrumented vehicle to collect naturalistic driving data. Data were collected on freeways in Maryland during times of moderate traffic flow. The instrumented vehicle was driven in a naturalistic way that was benign to the surrounding traffic. Tailgating events were detected using the empirical data and a model of safe following distance. We tested and affirmed the hypothesis that tailgaters of short tailgating duration are more willing to follow at close following distances than those who tailgated for longer durations. We also tested and affirmed the hypothesis that following vehicle speeds are strongly influenced by lead vehicle speeds. We studied the causal relations between certain observable data from the lead vehicle and possible reactions in the following vehicle. We contributed new estimates of driver reaction times, focusing on a subset of the population deemed to be tailgating at the time. We also conducted a new calibration of the well-known GHR car-following model that is specific to tailgating situations. The data and method for collecting the data are contemporary and relevant to current modes of thinking in traffic flow theory. The results can contribute directly to models and parameter estimates in microscopic simulators. Many of the results would also be of use in the automotive industry, for the development of driver safety assistance systems and countermeasures. Finally, we think the results could be useful for driving instructors, to help students understand better this dangerous driving behavior. In the end, we hope that this study could help to improve traffic safety by reducing the number of crashes resulting from this behavior.