The National Highway Traffic Safety Administration (NHTSA) and the California Department of Motor Vehicles (CA DMV) collect and utilize data from traffic accidents caused by Automated Driving Systems (ADS) driving on real roads, as a policy. Leading autonomous driving technology companies such as Tesla and Waymo collect their own driving and accident data and use them for technology advancement. ADS traffic accident data that occur when driving on real roads are valuable for identifying problems in unexpected situations. This study analyzes the risk of traffic accidents by Operational Design Domain (ODD) on ADS traffic accident data that occurred while driving on an actual road and aims to present a road traffic law-based driving ability evaluation scenario in a complex ODD configuration in high-risk situations, wherein an ADS can be particularly vulnerable in mixed traffic situations. The actual road traffic accident data of ADS from 2,289 accidents as provided by the NHTSA were analyzed. Analysis of the characteristics of ADS traffic accidents revealed that accidents occurred mainly on ordinary ODDs with high traffic demand during actual road driving, that is, on dry roads during clear days and daylight. In traffic situations including ADS and Human Driving Vehicle(HDV), approximately 40% of traffic accidents were confirmed to have occurred because of HDV colliding with stationary ADS and occurred in unexpected situations, such as changing the HDV when driving straight ahead of the ADS. Results of analyzing the risk of traffic accidents on the driving status of ADS by ODD, showed that the risk of traffic accidents that occurred while the ADS was driving straight ahead was 2.27, with dry road conditions, sunny weather, and a road speed limit of 21 to 30 mph at night when streetlights were turned on. Thus, the ADS road traffic law-based driving ability evaluation scenario can be used to evaluate whether to recognize and respond to accident risk situations by developing ADS road traffic law-based driving ability evaluation scenarios for situations vulnerable to accidents due to HDV cut-in in traffic situations that include ADS and HDV. In future, this can be used as basic data for preparing related regulations and institutional devices, such as traffic accident investigations and driving ability evaluations by ADS.

This study analyzes and compares the determinants of accident severity between human-driven vehicle (HDV) and autonomous vehicle (AV) mixed environments using collision data from the California Department of Motor Vehicles . To address the high dimensionality and categorical complexity of the dataset, an XGBoost-based classification model was developed and the Shapley additive explanations method was employed to explain the contribution and directional influence of each explanatory variable. An undersampling and ensemble approach was utilized to mitigate class imbalances and enhance the model stability. The results revealed that in an HDV environment, driver perception and evasive responses were dominant factors influencing crash outcomes, with collision direction and relative speed significantly affecting the severity. By contrast, in the AV–HDV mixed environment, intersection conditions and complex driving contexts were associated with higher accident severity, thus demonstrating the current limitations of AV systems in managing unstructured traffic scenarios. These findings suggest that as AV deployment progresses, the key determinants of crash severity shift from human behavioral factors to system and environmental factors, thus providing empirical insights for future AV safety evaluations and policy frameworks.

With a view towards reducing traffic accidents on roadways, various methods have been considered to predict accidents. In this study, we analyze traffic accident frequency models that employ fixed- and random-parameter negative binomial approaches. Random parameters enable the inclusion of unobserved heterogeneity in traffic accident data, which current popular methods with fixed parameters such as Poisson or negative binomial models cannot consider in terms of time variation or segment-specific effects. A continuous, unbalanced panel of accident histories for 208 four-way signalized intersections on national highways in Seoul was used to estimate a traffic accident occurrence model that considered traffic volumes and various geometric characteristics at intersections. The results revealed that the left-turn exclusive lanes and traffic volumes on minor roads had random parameters that affected the likelihood of accident frequencies differently; the other variables were found to significantly affect traffic safety at the intersections on the national highways as fixed parameters. Based on these results, it can be concluded that the same traffic safety facilities have different effects on traffic accidents on major and minor roads. The insights from this study suggest the need for a broader analysis of integrated guidelines for facilities that impact intersection accident propensities.

The primary objective of the study is to analyze and evaluate the situation and trends of inland waterway traffic accidents in Vietnam from 2017 to 2024. The study employs reliable secondary data sources, which are analyzed using statistical methods and heatmap applications to examine and assess trends in inland waterway traffic accidents in Vietnam. The results indicate a steady increase in both the number and scale of inland waterway accidents nationwide over the years. Additionally, the accident-prone areas in key inland waterways will be identified. Based on these findings, the research team has proposed recommendations and solutions aimed at improving traffic safety on Vietnam's inland waterways.

This study analyzed actual traffic accident data to select humans’ unavoidable accidents and to examine whether avoidance is possible after AEBS(Advanced Emergency Braking System) is applied to these accidents. In cases where avoidance is not possible with AEBS, those accidents were determined to be examples where V2X(Vehicle-to-Everything) technology is necessary. Subsequently, by applying V2V(Vehicle-to-Vehicle) and V2I(Vehicle-to-Infrastructure) communication technologies, this research analyzed the possibility of accident avoidance. The results confirmed that the application of V2X technology enables accident avoidance. Additionally, by applying various variables, it identified limitation scenarios that cannot be resolved by V2X technology, and discussed strategies for accident avoidance in such situations.

PURPOSES : This study aims to identify the thresholds at which various factors affecting traffic crashes lead to actual traffic crashes METHODS : To verify the thresholds, we created scenarios and ran simulations with a combination of factors that affect traffic crashes. Lateral offset and minimum TTC were used to evaluate whether an incident occurred. RESULTS : In the first scenario, the most significant factor affecting traffic crashes is curvature, and it was found that the smaller the curvature(200 meters or less), the greater the deviation from the lane. And in the second scenario, especially the passenger car scenario, no accidents occurred when the curvature was greater than 90 meters and the speed was 40 km/h or less. The smaller the curvature and the higher the speed, the more accidents occurred. Similarly, in the bus scenario, no accidents occurred when the curvature was 120 meters or more and the speed was 30 km/h or less. Also, accidents tended to occur when the curvature was smaller and the speed was higher. CONCLUSIONS : Through this study, we derived the thresholds of factors that influence traffic crashes. The results are expected to help design and operate roads in the future and contribute to reducing traffic crashes.

목적 : 본 연구는 미국의 조건부 운전면허 관련 상세 운영 현황을 알아보고, 분석한 결과를 바탕으로 고령운전자 안전운전 을 위한 국내 조건부 운전면허 도입 시 참고가 될 수 있는 기초자료를 마련하고자 하였다. 또한, 운전재활전문가로서 작 업치료사의 조건부 운전면허에 대한 관심을 제고시키고자 하였다. 연구방법 : 미국자동차협회 교통안전재단(AAA Foundation for Traffic Safety)의 자료를 참고하여 미국의 고령운전자 사고율이 25% 이상에 해당하는 상위 4개 주인 켄터키, 미시시피, 몬태나, 와이오밍 주의 고령운전자 조건부 운전면허 관 련 현황을 정리하였다. 결과 : 본 연구에서는 미국 4개 주의 고령운전자의 운전면허 갱신, 고령운전자 조건부 운전면허 실시 및 제한 방법, 고령 운전자 조건부 운전면허 결정 판단 주체, 지역 현지 심사관(Local examiner)의 고령운전자 조건부 운전면허 결정과 관 련된 현황을 주로 파악해보았다. 조건부 운전면허 제한 유형으로는 고속도로 접근 제한과 거리 제한, 차량 장비 등이 있 었고 각 주의 공통적인 부분으로는 주간이나 낮에만 운전할 수 있도록 야간운전 제한 및 속도를 제한하고 있음을 파악하 였다. 결론 : 본 연구는 미국 4개 주의 고령운전자 조건부 운전면허제도 관련 현황 분석을 통해 우리나라 고령운전자 조건부 운 전면허제도 도입 시 참고 가능한 자료임에 의의가 있다. 앞으로 우리나라 고령운전자를 위한 국내 환경에 맞는 조건부 운전면허제도 도입과 관련된 기초자료로 활용될 수 있을 것으로 사료된다.

PURPOSES : There are significant differences in traffic accident rates depending on various road conditions and environments. However, the current traffic accident rates on national highways are classified relatively simply, and it is also difficult to accurately calculate the crash modification factor. Therefore, this study aimed to improve the traffic accident rates on national highways by presenting an algorithm for categorizing the traffic accident rates of national highway into four types (older and modern roads, and urban and rural roads). METHODS : The problems in the current rate of traffic accidents were derived, Traffic accident analysis system(TAAS) was used for the traffic accident data, and the road traffic volume statistical yearbook was used for the traffic volume data. After dividing the national highways into older and modern roads and urban and rural roads, the rates of traffic accidents were calculated and compared with the current accident rates. RESULTS : The accident rate of modern roads was found to be lower than that of older roads, and was lower in rural areas than in urban areas. From comparing the results of this study with Korea development institute(KDI) guidelines, older roads and urban roads exceeded the value in the KDI guideline, whereas the rates of modern roads and rural roads were lower than the KDI value. CONCLUSIONS : The accident rate accuracy was improved by subdividing the accident rates into four types. Therefore, it is expected that the accuracy and reliability of economic analysis on road projects will be improved.

해양사고 예방을 위해서는 사고의 원인과 결과에 대한 분석 및 진단뿐만 아니라, 사고의 발생 패턴과 변화 추이를 예측함으로 써 정량적 위험도를 제시할 필요성이 있다. 선박교통과 관련된 해양사고 예측은 선박의 충돌위험도 분석 및 항해 경로 탐색 등 선박교통 의 흐름에 관한 연구가 주로 수행되었으며, 해양사고의 발생 패턴에 대한 분석은 전통적인 통계 분석에 따라 제시되었다. 본 연구에서는 해양사고 통계 자료 중 선박교통관련 사고의 월별, 시간대별 발생 현황 데이터를 활용하여 해양사고 발생 예측 모델을 제시하고자 한다. 국내 해양사고 발생 현황 중 월별, 시간대별 데이터 집계가 가능한 1998년부터 2021년까지의 통계자료 중 선박교통 관련 데이터를 분류하 여 정형 시계열 데이터로 변환하였으며, 대표적인 인공지능 모델인 순환 신경망 기반 장단기 기억 신경망을 통하여 예측 모델을 구축하 였다. 검증데이터를 통하여 모델의 성능을 검증한 결과 RMSE는 초기 신경망 모델에서 월별 52.5471, 시간대별 126.5893으로 나타났으며, 관측값으로 신경망 모델을 업데이트한 결과 RMSE는 월별 31.3680, 시간대별 36.3967로 개선되었다. 본 연구에서 제안한 신경망 모델을 기 반으로 다양한 해양사고의 특징 데이터를 학습하여 해양사고 발생 패턴을 예측할 수 있을 것이다. 향후 해양사고 발생 위험의 정량적 제 시와 지역기반의 위험지도 개발 등에 관한 추가 연구가 필요하다.

PURPOSES : In this study, the factors affecting the severity of traffic accidents in highway tunnel sections were analyzed. The main lines of the highway and tunnel sections were compared, and factors affecting the severity of accidents were derived for each tunnel section, such as the tunnel access zone and tunnel inner zone. METHODS : An ordered probit model (OPM) was employed to estimate the factors affecting accident severity. The accident grade, which indicates the severity of highway traffic accidents, was set as the dependent variable. In addition, human, environmental, road condition, accident, and tunnel factors were collected and set as independent variables of the model. Marginal effects were examined to analyze how the derived influential factors affected the severity of each accident. RESULTS : As a result of the OPM analysis, accident factors were found to be influential in increasing the seriousness of the accident in all sections. Environmental factors, road conditions, and accident factors were identified as the main influential factors in the tunnel access zone. In contrast, accident and tunnel factors in the tunnel inner zone were found to be the influencing factors. In particular, it was found that serious accidents (A, B) occurred in all sections when a rollover accident occurred. CONCLUSIONS : This study confirmed that the influencing factors and the probability of accident occurrence differed between the tunnel access zone and inner zone. Most importantly, when the vehicle was overturned after the accident occurred, the results of the influencing factors were different. Therefore, the results can be used as a reference for establishing safety management strategies for tunnels or underground roads.

This study intends to present a traffic node-based and link-based accident prediction models using XGBoost which is very excellent in performance among machine learning models, and to develop those models with sustainability and scalability. Also, we intend to present those models which predict the number of annual traffic accidents based on road types, weather conditions, and traffic information using XGBoost. To this end, data sets were constructed by collecting and preprocessing traffic accident information, road information, weather information, and traffic information. The SHAP method was used to identify the variables affecting the number of traffic accidents. The five main variables of the traffic node-based accident prediction model were snow cover, precipitation, the number of entering lanes and connected links, and slow speed. Otherwise, those of the traffic link-based accident prediction model were snow cover, precipitation, the number of lanes, road length, and slow speed. As the evaluation results of those models, the RMSE values of those models were each 0.2035 and 0.2107. In this study, only data from Sejong City were used to our models, but ours can be applied to all regions where traffic nodes and links are constructed. Therefore, our prediction models can be extended to a wider range.

PURPOSES : For vehicle-alone accidents with a high mortality rate, it is necessary to analyze the factors influencing the severity of roadside fixed-object traffic accidents. METHODS : A total of 313 roadside fixed obstacle traffic accidents, variables related to fixed obstacles, and variables related to road geometry were collected. The estimation model was constructed with data collected using an ordinal probit regression model. RESULTS : Piers, vertical slopes, and distances between roads and objects were the primary causes of increased accident severity. CONCLUSIONS : Countermeasures, such as object removal, relocation, clear zones, frangibles, breakaway poles, etc., are required for accident-prone or dangerous points.

PURPOSES : The purpose of this study is to compare applicability, explanation power, and flexibility of traffic accident models between estimating model using the statistical method and the machine learning method. METHODS: In order to compare and analyze traffic accident models between model estimated using the statistical method and machine learning method, data acquisition was conducted, and traffic accident models were estimated using statistical methods such as negative binomial regression model, and machine learning methods such as a generalized regression neural network (GRNN). Then, the fitness of model as R2, root mean square error (RMSE), mean absolute percentage error (MAPE), accuracy, etc., were determined to compare the traffic accident models. RESULTS: The results showed that the annual average daily traffic (AADT), speed limits, number of lanes, land usage, exclusive right turn lanes, and front signals were significant for both traffic accident models. The GRNN model of total traffic accidents had been better statistical significant with R2: 0.829, RMSE: 2.495, MAPE: 32.158, and Accuracy: 66.761 compared with the negative binomial regression model with R2: 0.363, RMSE: 9.033, MAPE: 68.987, and Accuracy: 8.807. The GRNN model of injury traffic accidents also showed similar results of model’s statistical significance. CONCLUSIONS: Traffic accident models estimated with GRNN had better statistical significance compared with models estimated with statistical methods such as negative binomial regression model.