1372247 - Mapping Location-Based Driving Volatility and Road Safety: A Study Using Large-Scale Crowdsourced Vehicle Movement Data

Research on road safety has long relied on crash or incident data to investigate the safety performances of roadway segments and intersections and to identify factors contributing to traffic crashes. With the advances in information technologies, massive volume of crowdsourced vehicle movement or trajectory data is being generated by mobile devices and electronic sensors in connected vehicles equipped with satellite navigation systems, which presents an opportunity for researchers and agencies to extend the understanding of road safety and its correlates. This study will take advantage of large-scale crowdsourced vehicle movement data that cover over 95% of roadways in Alabama to explore the Location-based Driving Volatility, defined as the percentage of extreme maneuvers at a location, in the road network. The vehicle instantaneous speeds and accelerations will be calculated to measure the driving volatility. At a location, a higher volatility value means a larger portion of vehicle maneuvers considered as extreme maneuvers (i.e., high accelerations/decelerations) at this location. The data used in this study are accurate to within a 3-meter radius and are highly granular data captured every 3 seconds for precise detail. The objective of this study is to map the location-based driving volatility on Alabama highways, and to link driving volatility to the safety outcomes (e.g., crash rates and injury severity) across the diverse roadway environments and socioeconomic contexts. This study will develop models to examine the spatial correlations between driving volatility and the safety outcomes. The study results are expected to identify roadway environments in four categories including 1) locations with high driving volatility and high crash rates, 2) locations with high volatility and low crash rates, 3) locations with low driving vitality and high crash rates, and 4) locations with low volatility and low crash rates. Different safety improvement strategies and countermeasures can be developed for four different roadway environments. Through exploring large-scale crowdsourced vehicle movement data, this study will contribute to the advancement of knowledge in the field of road safety that has been traditionally relied on crash or incident data.