Developing a Comprehensive Vulnerable Road User Safety Screening Method Using Multi-Level Data

Abstract

Despite significant advancements in motor vehicle safety, the number of fatalities resulting from road crashes has been increasing in recent years. The resultant injuries from roadway crashes are more severe for vulnerable road users (VRUs), that is, pedestrians and cyclists. High disparity is observed in the distribution of the burden of these crashes across areas and ethnic and race groups; therefore Federal Highway Administration (FHWA) requires states to incorporate equity-related variables in the assessment of VRU safety. Conventionally, the Highway Safety Manual (HSM) provides methodologies for identifying locations with high risks of crashes, but those methods have limited applications in prioritizing low-volume sites with a higher crash rate, as in the case of VRUs. This study first explores a wide range of equity-related variables at an area level in addition to site characteristics of state-maintained signalized intersections in Connecticut. This study also proposed a framework to combine site-level and area-level risk analysis to generate an aggregate network screening result, using a Bayesian Thurstonian model and a mean score ranking method. The proposed data-driven safety assessment framework is a novel practice in transportation safety analysis and shows promising results for network screening based on systemic safety principles.