Development of Integrated Driving Evaluation Index by Proportion of Autonomous Vehicles for Future Intelligent Transportation Systems

As the market penetration rate (MPR) of autonomous vehicles increases, it is expected that the safety of mixed traffic situations will change due to interactions between vehicles. A proactive safety analysis of mixed traffic situations is needed for future intelligent transportation systems; thus, it is necessary to determine the driving safety evaluation indicators that have a significant impact on identifying hazardous sections of actual roads by each MPR. The purpose of this study is to simulate autonomous vehicle behavior by analyzing real-world autonomous vehicle data and to derive a promising integrated driving safety evaluation index for mixed traffic. Autonomous vehicle driving data from an autonomous mobility testbed in Seoul were collected and analyzed to assess autonomous vehicle behavior in VISSIM. The simulation environment was established to match the real road environment. Decision tree (DT) analysis was adopted to derive the indicators influencing the classification of hazardous sections of real roads by MPR. The vehicle–vehicle interaction indicators used to evaluate driving safety were applied as the input variables of the DT, and the classification of real-world hazardous road sections was the output variable. An integrated evaluation index was developed using the promising evaluation indicators and information gains derived for each MPR. The most hazardous section and the factors affecting the driving safety of the section based on the integrated evaluation index for each MPR were then presented. The results of this study can be utilized to proactively identify hazardous road sections in the real world through simulations of mixed traffic conditions.