Analysis of the Impact of Driver Behavior Models on Performance of Forward Collision Warning Systems

A critical aspect of vehicle safety analysis is understanding the impact of human behavior on the overall performance of the system. Since active safety systems (such as collision warning systems) are designed for escaping a life threatening situation, evaluation and validation of their components and the underlying algorithms are often costly, and cannot be achieved without direct exposure of test subjects to hazards. As an alternative, researchers try to use simulation tools to design and evaluate their safety algorithms prior to building test-beds for validation. This study proposes a comprehensive simulation framework for vehicular active safety system analysis. It comprises different domains of communication, transportation, and vehicular safety. The simulation framework supplies a driver model and allows for analysis of human reaction and behavior to an imminent warning. Modeling human behavior is specially challenging and requires an in-depth breakdown and investigation of the class of actions a human driver may select from. In this study, we look at the primary form of a driver reaction, i.e. braking when a warning is issued. We investigate different levels of braking and complete our analysis by studying the effect of each braking level on a subject forward collision warning algorithm.