Methods and Metrics for Evaluation of an Automated Real-Time Driver Warning System

Abstract

There is significant interest among traffic management personnel in the use of automated warning systems to provide drivers with real-time information on hazardous conditions related to traffic, limited visibility, or roadway obstructions. However, the effectiveness of such systems in achieving desired traffic safety improvements has not yet been well quantified. Relative influences on traffic safety can be assessed in many ways, and overall conclusions must be based on an appropriate set of metrics and methodologies for a particular implementation. This work, supported by the California Office of Traffic Safety and the California Department of Transportation (Caltrans), builds on prior research to develop, deploy, and test various metrics and methods to evaluate a large-scale real-time driver warning system, the Caltrans automated warning system, installed on Interstate 5 and State Route 120 near Stockton, California. Methods include the analysis of historical accident data over an 11-year period, a direct assessment of the operational behavior of the system correlated with accident data, and a study of the direct effects of real-time warning messages on driver behavior. Instrumentation deployed to facilitate these detailed analyses is described. The resultant body of data supports the correlation of measurable traffic flow parameters with relative traffic safety.