Towards Collision Prediction and Control: Modeling Longitudinal and Lateral Driver Behavior Leading to Realistic Collision Rates Open Access
Downloadable ContentDownload PDF
With millions of traffic collisions occurring each year and tens of thousands of these collisions resulting in fatalities, the need to keep improving traffic safety is apparent. One way to accomplish this is to look into traffic simulation models in order to test different traffic situations and variables. As of today, there are models that allow for acceleration and lane-changing simulations for traffic conditions before a collision occurs. In addition, there are crash reconstruction and vehicle dynamic simulation models for after a collision occurs. However, there is no current linkage between these pre- and post-collision models. This linkage is necessary in order to understand driver behavior and vehicular movements that occur before, during, and after a collision occurs. By bridging the gap between pre- and post-collision traffic models, the ability to identify dangerous situations that could lead to a collision formation becomes a reality. This research outlines the various types of microscopic traffic models in a literature review, presents a genealogy of the models, selects the models with the best potential to be linked, and prepares the implementation for the linkage. Additionally, this document shows the preliminary lateral trajectory results for the adopted lane-changing model and data analysis on crash involvement rates produced by the adopted acceleration model.