An Integrated Automated Driving Risk Indicator in Urban Mixed Traffic Environments

In this study, a novel methodology is proposed to evaluate automated driving safety in mixed traffic environments, including autonomous vehicles (AVs) and manually driven vehicles (MVs). An open-source AV dataset obtained from a real-world autonomous mobility testbed in Korea was used for methodology development and evaluations. The driving behavior was evaluated using well-known promising indicators, including the standard deviation of the vehicle speed, acceleration noise, standard deviation of the lane offset, time to collision (TTC), and deceleration to avoid a crash (DRAC). Min-max and max-min normalization was performed to unify the units of the evaluation indicators. The importance of each driving safety indicator was derived through the Analytical Hierarchy Process (AHP) performed by traffic experts, and the weights were estimated based on the average of the collected importance. The normalized indicators were integrated to obtain the automated driving risk score (ADRS), which is regarded as a measure of automated driving safety. The automated driving safety degraded considerably in road sections where right turns were made at intersections and that had a bus stop. Hazardous driving events of AVs were visualized, which is useful for monitoring mixed traffic safety and developing effective countermeasures for proactive road safety management.