Leveraging connected vehicle data for roadway safety evaluation and management: segment- and intersection-level driving volatility analysis

Abstract

Abstract The advancement of Intelligent Transportation Systems (ITS) and the proliferation of Connected and Automated Vehicles (CAVs) have enabled the collection of large-scale, high-resolution vehicle movement data through crowdsourcing. This wealth of data presents a unique opportunity to move beyond traditional, retrospective crash-based analyses toward more proactive and dynamic approaches to traffic safety assessment and management. This study proposes a comprehensive framework for driving volatility analysis at both the segment and intersection levels using large-scale connected vehicle trajectory data. Driving volatility—characterized by abrupt changes in acceleration, deceleration, and speed—is used as a behavioral safety surrogate to identify locations with elevated crash risk before crashes occur. The analysis integrates volatility metrics with roadway geometry, traffic volume, and historical crash data to develop predictive models of crash frequency. Results reveal a strong and statistically significant relationship between elevated driving volatility and higher crash frequencies, with segment-level volatility demonstrating particularly robust predictive power. These findings highlight the potential of incorporating driving volatility into ITS-based safety monitoring systems, enabling real-time detection of high-risk locations and more targeted, data-driven safety interventions. By leveraging CAVs as mobile sensors within an ITS framework, transportation agencies can enhance situational awareness, support proactive traffic safety management strategies, and improve roadway safety outcomes. This approach represents a significant shift toward real-time, behavior-based safety diagnostics in the era of intelligent and connected transportation systems.