As the number of autonomous vehicles increases, the number of accidents also increases every year. These incidents include general car/traffic accidents and may introduce new potential issues such as cybersecurity and sensor errors of autonomous vehicles. The existing traffic accident investigation method has limitations in identifying the cause of the autonomous vehicle accident. Some states in the US (e.g., California and Texas) introduced limited items of Lv. 2 autonomous vehicle accidents. For instance, “vehicle level” and “autonomous mode/conventional mode” are being investigated to identify the cause of autonomous vehicle accidents. Therefore, it is crucial to propose accident investigation items and procedures in preparation for various autonomous vehicles that may occur in the future. In order to address these issues, this study collected reports used in existing traffic accident investigations, autonomous driving‐related reports and literature, and accident videos involving autonomous driving to build investigation items. First, we reviewed the items required for investigation in the event of a conventional vehicle accident and added additional investigation items deemed necessary to be reviewed in addition to the existing reports. Second, based on the conventional vehicle accident investigation items, this study derived the autonomous driving traffic accident investigation items. Finally, an accident involving autonomous vehicle(s) investigation process was established that can be used by the police and various investigation jurisdictions. The results of this paper can improve the understanding of the cause of future traffic accidents involving autonomous vehicles.

Traffic accident investigation has been used to reconstruct the cause and condition of the accident by using images and driving record data that introduce IT technology, and to analyze both sides of the accident, skid marks, and vehicle damage to identify the perpetrators and victims. However, level 3 self-driving vehicles are the most important factor in determining the cause and imputation of the accident by the driver or manufacturer with control information at the time of the accident. It is also developing into a network and connected vehicle with various restrictions such as the burden of the price of sensors equipped with self-driving cars and climate and rapidly changing road traffic information. In addition, network and connected vehicle driving data are stored on the outside, or various devices and sensors are installed to store information on the outside for convenience in operation, and efforts to enact laws are continuing. This paper attempts to propose a traffic investigation digital framework using digital data generated by these devices and sensors.

Recently, the commercialization of autonomous vehicles has increased the importance of verifying vehicle safety through autonomous trials. Autonomous driving trials are conducted in limited areas within artificially constructed test roads and pilot districts and directly explore road sections and areas with similar environments to ensure the safety of AVs driving on real roads. Many previous studies have evaluated the complex response potential of AVs by deriving edge scenarios to ensure their safety. However, difficulties arise in exploring real roads with traffic accident factors and configurations similar to those in edge scenarios, making validation on real roads uncertain. This paper proposes a novel method for exploring pilot zones using traffic accident data to verify the safety of autonomous vehicles (AVs). The method employs a CNN + BiGRU model trained on DMV dataset data to classify traffic accidents as AV- or human-caused. The model’s classification accuracy was evaluated using recall, precision, F1 score, and accuracy, achieving 100.0%, 97.8%, 98.9, and 99.5%, respectively. The trained model was applied to the KNPA dataset, identifying 562 out of 798 cases as AV-like, indicating potential areas of high accident density due to AV operation. Outlier detection and DBSCAN clustering were utilized to identify compact pilot zones, effectively reducing the area size compared to raw data clusters. This approach significantly lowers the cost and time associated with selecting test roads and provides a viable alternative for countries lacking real AV accident data. The proposed method’s effectiveness in identifying pilot zones demonstrates its potential for advancing AV safety validation.

Traffic accidents are widely recognized as being a significant societal disaster that causes severe human, economic, and social damage. While various factors, such as environmental, road, and human factors are commonly associated with traffic accidents, this study proposes a novel approach to consider the directionality of traffic accidents. It examines its impact on traffic safety in Chungcheongnam-do province, South Korea. Using police reports, we estimated the directionality of each accident by calculating the angle between its occurrence site and destination. Our experimental results revealed a statistically significant association between the directionality of traffic accidents and their incidence rates in Chungcheongnam-do province, with a higher frequency of accidents occurring in a Southbound direction. These findings suggest that considering the directionality of traffic accidents could enhance the effectiveness of traffic safety policies and interventions. Further research is needed to explore this approach’s underlying mechanisms and potential applications. Nevertheless, this study provides important insights for policy makers and practitioners to develop targeted and practical measures for improving traffic safety.