A holistic approach for proactive safety assessments of road traffic employing crowdsourcing and vehicle kinematics

Traditional methods of traffic safety assessment rely primarily on crash data. This presents several inherent limitations, including the need for incidents to occur before analysis and the underreporting of crashes. To address these challenges, this study introduces a comprehensive methodology that integrates crowdsourced reports on danger spots, human factor capturing, and vehicle telematics-based hard-braking events with crash data to proactively assess the safety of road networks. By incorporating road users’ perceptions and experiences of safety deficits associated with the built infrastructure, the methodology provides valuable insights that complement traditional data sources. Crowdsourced danger spots and clusters of hard-braking events are validated through on-site audits to obtain objective safety ratings independently of crash data. A statistical analysis using a logistic regression model reveals that both crowdsourced danger spots and vehicle kinematics are statistically discernible factors for predicting safety ratings. Our results indicate a disproportionate representation of individual road user groups in crash data. This highlights the value of human-factor insights through crowdsourced data and vehicle telematics-based hard-braking events in augmenting crash-based analysis. By combining all three data sources, authorities can identify high-risk locations with enhanced accuracy, facilitating the development of targeted countermeasures to improve overall traffic safety.