Hazardous Driving Event Detection and Analysis System in Vehicular Networks

Abstract

This study proposes a novel highway traffic surveillance system that is capable of detecting hazardous driving maneuvers through the use of an in-vehicle sensor and transmission of the event data detected to a traffic management center through vehicle-to-vehicle and vehicle-to-infrastructure wireless communication; this system is referred to as the “hazardous driving event detection and analysis system in vehicular networks.” Compared with existing surveillance systems, the main novel feature of the proposed system lies in its ability to detect hazardous driving maneuvers that have the potential to lead to crashes. Three major components of the system are introduced in this study: an algorithm for detecting hazardous driving events, a method for deriving a new index incorporating expert judgment for evaluation of the risk level of the traffic stream on the basis of analyses of hazardous events that are detected, and field implementation of the proposed system in a test bed with real-time and historical data. Extensive field tests were conducted in the test bed to fine-tune the prototypical system. The methodology and field implementation presented in this study have potential value to help highway traffic agencies monitor and evaluate traffic streams with a focus on traffic safety. The proposed system is expected to be effective as support for the development of various traffic information control strategies to enhance traffic safety on highways.