Accident; analysis and prevention最新文献

{"title":"Opposing-through crash risk forecasting using artificial intelligence-based video analytics for real-time application: integrating generalized extreme value theory and time series forecasting models","authors":"Md Mohasin Howlader,&nbsp;Md Mazharul Haque","doi":"10.1016/j.aap.2025.108073","DOIUrl":"10.1016/j.aap.2025.108073","url":null,"abstract":"<div><div>Recent advancements in artificial intelligence (AI) and traffic sensing technologies provide significant opportunities for real-time crash risk forecasting. While forecasting based on historical crash data yields macroscopic insights into future crash risks, such information is often insufficient for real-time applications. In contrast, traffic conflict techniques (TCTs) leveraged by extreme value theory (EVT) and AI-based video analytics have enabled crash risk estimation to a granular level, presenting a promising potential for real-time applications. This study develops a unified framework of integrating generalized extreme value (GEV) theory with parametric and non-parametric forecasting models to predict opposing-through crash risks at signalized intersections. A deep neural network-based computer vision technique was employed to extract post encroachment time (PET) traffic conflicts from 97 h of video footage. Crash risks were estimated using a non-stationary GEV model, incorporating traffic conflict counts, speed variations, and signal timing characteristics. These risk estimates were then forecasted using autoregressive integrated moving average (ARIMA), gated recurrent unit (GRU), and long short-term memory (LSTM) models to analyze short-term crash trends. Results show that the mean crash frequency estimates fell within the 95 % confidence limits of observed crashes and confirm the adequacy of the developed EVT model in estimating opposing-through crashes. The autoregressive and recurrent neural network models exhibit similar forecasting accuracy for crash risk forecasting, with reliable predictions extending up to 11 future signal cycles. The proposed real-time crash risk forecasting framework can be a crucial component of an intelligent transport system, leading to proactive safety management for signalized intersections.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"218 ","pages":"Article 108073"},"PeriodicalIF":5.7,"publicationDate":"2025-05-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143913152","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Spatiotemporal urban traffic safety analytical framework by integrating nonparametric approaches","authors":"Youngwoong Kim ,&nbsp;Dongwoo Lee ,&nbsp;Sybil Derrible","doi":"10.1016/j.aap.2025.108088","DOIUrl":"10.1016/j.aap.2025.108088","url":null,"abstract":"<div><div>Since more than 75% of the population lives in cities, it is crucial to create a safe transportation environment for all urban residents. In this context, significant efforts are required to mitigate potential accident risks and make cities more inclusive. To gain insights into an inclusive traffic safety environment and develop a system that provides useful traffic safety information accessible to all stakeholders, from end-users to decision-makers, this article aims to develop a novel nonparametric modeling framework, the Mixed-Effect Tree Ensemble with a Gaussian Process (ME-GP), for city-wide traffic safety analysis.<!--> <!-->In this study, we use police-reported accident data from Seoul (South Korea). The framework leverages the advantages of integrating nonparametric modeling approaches to predict accident risks at the road-segment level while accounting for spatiotemporal heterogeneity and unobserved data complexities. The Gaussian process, in particular, enables us to capture nonlinearities and discontinuities when estimating random parameters. Due to the nature of the police-reported accident data, Tree-ensemble is integrated with the Gaussian process. Compared to other nonparametric models, including integrated modeling approaches, ME-GP demonstrated a 15% improvement in predictive accuracy and lower variance in out-of-sample predictions, highlighting its robustness and reliability. The result revealed that demographics, traffic conditions, and road structure are the most determinant factors in accident risks. As expected, the relationship between determinant factors and accident risks is nonlinear and spatiotemporally heterogeneous. Elderly accidents were found to have a maximum accident risk of 20% higher than that of youth. In contrast, children who are also physically vulnerable showed a lower accident risk, which is partly because of school zones that effectively protect children. The findings from the framework can provide useful insights into establishing safe and inclusive urban networks.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"218 ","pages":"Article 108088"},"PeriodicalIF":5.7,"publicationDate":"2025-05-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143913151","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Prioritizing safety funding using severity weighted risk scores","authors":"Grant G. Schultz ,&nbsp;Tomas Barriga Aristizabal ,&nbsp;Jace Ritchie ,&nbsp;Richard L. Warr","doi":"10.1016/j.aap.2025.108083","DOIUrl":"10.1016/j.aap.2025.108083","url":null,"abstract":"<div><div>Limited funding availability requires government agencies to focus transportation funding on locations most in need of safety improvements. The Two-Output Model for Safety (TOMS) was created to prioritize segments and intersections for safety analysis. The TOMS compiles input data, prepares the data files so that the format and content are consistent, and then two different processes occur. The first is to segment roadways based on five variables: average annual daily traffic, functional class, lanes, speed limit, and urban code. The second is to assign the physical characteristics of the roadway to each individual intersection. TOMS outputs a segment file and an intersection file used for statistical analysis and are the “two outputs” referenced by the name of the model. The segments and intersections are analyzed using severity and total number of crashes at the sites. An excess weighted risk score was developed using an equivalent property damage only value to analyze the severity and number of crashes concurrently. The segments and intersections with the highest excess weighted risk scores are prioritized as locations for safety funding. A report compiler is then executed to create two-page safety reports that contain roadway and crash information organized in a manner that allows governing agencies to identify how many crashes are occurring at a site and the manner of collision for the crashes. The research presented in this paper shows that the simultaneous use of intersection and segment analysis combined with excess weighted risk scores can provide insight into the prioritization of safety funding.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"218 ","pages":"Article 108083"},"PeriodicalIF":5.7,"publicationDate":"2025-05-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143916710","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Artificial intelligence automated solution for hazard annotation and eye tracking in a simulated environment","authors":"Piyush Pawar ,&nbsp;Benjamin McManus ,&nbsp;Thomas Anthony ,&nbsp;Jingzhen Yang ,&nbsp;Thomas Kerwin ,&nbsp;Despina Stavrinos","doi":"10.1016/j.aap.2025.108075","DOIUrl":"10.1016/j.aap.2025.108075","url":null,"abstract":"<div><div>High-fidelity simulators and sensors are commonly used in research to create immersive environments for studying real-world problems. This setup records detailed data, generating large datasets. In driving research, a full-scale car model repurposed as a driving simulator allows human subjects to navigate realistic driving scenarios. Data from these experiments are collected in raw form, requiring extensive manual annotation of roadway elements such as hazards and distractions. This process is often time-consuming, labor-intensive, and repetitive, causing delays in research progress.</div><div>This paper proposes an AI-driven solution to automate these tasks, enabling researchers to focus on analysis and advance their studies efficiently. The solution builds on previous driving behavior research using a high-fidelity full-cab simulator equipped with gaze-tracking cameras. It extends the capabilities of the earlier system described in Pawar’s (2021) “Hazard Detection in Driving Simulation using Deep Learning”, which performed only hazard detection. The enhanced system now integrates both hazard annotation and gaze-tracking data.</div><div>By combining vehicle handling parameters with drivers’ visual attention data, the proposed method provides a unified, detailed view of participants’ driving behavior across various simulated scenarios. This approach streamlines data analysis, accelerates research timelines, and enhances understanding of driving behavior.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"218 ","pages":"Article 108075"},"PeriodicalIF":5.7,"publicationDate":"2025-05-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143916709","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Modeling motorcycle crash frequency on rural multilane segments in Kentucky","authors":"Bharat Kumar Pathivada,&nbsp;Arunabha Banerjee,&nbsp;Kirolos Haleem,&nbsp;Tathagatha Khan,&nbsp;Dylan Justice","doi":"10.1016/j.aap.2025.108085","DOIUrl":"10.1016/j.aap.2025.108085","url":null,"abstract":"<div><div>Despite motorcycle crashes accounting for a large percentage of traffic fatalities in the U.S., studies investigating motorcycle crash frequency are relatively limited. This study took the initiative and developed motorcycle crash-specific safety performance functions (SPFs) along rural multilane road segments in Kentucky, separately for the pre-COVID-19 pandemic (2015–2019) and post-COVID-19 pandemic (2020–2022) periods. Eight years of motorcycle crash records (2015 through 2022) and site-specific characteristics (e.g., shoulder width and annual average daily traffic “AADT”) were collected and used. Conway-Maxwell-Poisson (CMP) and heterogeneous Conway-Maxwell-Poisson (HTCMP) models were fitted and compared while accounting for motorcycle crash under-dispersion (i.e., when crash variance is less than its mean). The study results showed that, for both pre- and post-pandemic periods, the HTCMP models outperformed their CMP counterparts based on various goodness-of-fit measures (e.g., likelihood ratio test “LRT”, Akaike information criterion “AIC”, and McFadden pseudo R-squared) and prediction performance measures (i.e., mean absolute deviance “MAD” and mean square prediction error “MSPE”). From the developed SPFs, for the pre-pandemic period, the presence of horizontal curves and undivided roadways were significantly associated with increased motorcycle crash frequency along rural multilane segments, while in the post-pandemic period, wider right shoulders and higher AADT were significantly associated with increased motorcycle crash frequency. The predicted crash frequencies while applying the best-fit models (i.e., the HTCMP models) were then used to identify and rank high-crash rural multilane segments in Kentucky. Based on the study findings, several countermeasures were proposed to improve motorcyclists’ safety along Kentucky’s rural multilane segments, e.g., adding centerline grooved rumble strips along undivided rural multilane roadways.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"218 ","pages":"Article 108085"},"PeriodicalIF":5.7,"publicationDate":"2025-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143908146","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Impact of the construction area layout on road safety in urban work zones","authors":"Peng Liu ,&nbsp;Chengyi Zhang ,&nbsp;David Arditi ,&nbsp;Ayoola Olorunnishola","doi":"10.1016/j.aap.2025.108092","DOIUrl":"10.1016/j.aap.2025.108092","url":null,"abstract":"<div><div>Despite the crucial role that work zone configurations play in traffic safety, there is a limited understanding of work zone configuration from the perspective of construction area layout within urban work zones impacts overall safety. This study addresses this gap by examining the critical influence of construction area layout on lane-changing, driving behavior passing by the construction area, and driving stability. Construction area layout was reflected by the position of heavy equipment in this study. A driving simulator experiment was conducted with 26 participants (14 males and 12 females) to simulate real-world urban work zone scenarios and assess the impact of construction area layout on safety. The experiment also considered two key safety-related independent variables: driver gender and ambient light condition. To evaluate driver behaviors and identify safety–critical patterns, parametric survival modeling, Multivariate analysis of variance (MANOVA), and regression analysis were employed. The findings highlight the significant impact of the construction area: (1) the construction area layout whereby heavy equipment was positioned closer to the center of the work zone (Position 2) prompted drivers’ merging distance to be 14.6% longer, underscoring the importance of heavy equipment at the start of the work zone (Position 1), and (2) Position 2 enables drivers to pass the work zone with a higher speed to pass by the construction area. The driver gender and ambient light condition can also have a significant effect. For example, the increased longitudinal velocity was observed during nighttime, suggesting a need for enhanced visibility and speed control. Male drivers tend to pass by the construction area with a more stable longitudinal velocity than female drivers. These findings are significant for improving work zone safety through careful consideration of construction area layout design, enhanced ambient lighting condition, and strategic safety interventions for specific groups. These insights offer valuable guidance for improving safety and operational efficiency in urban work zones, reducing the risk of accidents, and safeguarding both drivers and construction personnel.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"218 ","pages":"Article 108092"},"PeriodicalIF":5.7,"publicationDate":"2025-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143906427","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Pattern recognition in crash clusters involving vehicles with advanced driving technologies","authors":"Reuben Tamakloe ,&nbsp;Mahdi Khorasani ,&nbsp;Subasish Das ,&nbsp;Inhi Kim","doi":"10.1016/j.aap.2025.108072","DOIUrl":"10.1016/j.aap.2025.108072","url":null,"abstract":"<div><div>Autonomous Vehicle (AV) technologies, including Advanced Driver Assistance Systems (ADAS) and Automated Driving Systems (ADS), have significant potential to reduce crashes caused by driver errors. However, as AVs become more prevalent on roadways, the number of crashes involving them is also increasing. While considerable research has explored factors contributing to AV crashes, a gap remains in understanding the critical risk factor patterns within clusters of ADAS- and ADS-engaged AV crashes. To address this gap, this study employs the Cluster Correspondence Analysis tool to cluster crash-related factors. The analysis identified three distinct clusters for both ADAS- and ADS-engaged AV crashes. For ADAS-engaged AVs, the most representative cluster involves fatal crashes at intersections, particularly those involving left-turning vehicles. In contrast, ADS-engaged AV crashes most commonly occur in daylight and involve non-motorists. Key differences were observed: when ADAS is engaged, rear-end crashes typically result in property damage only, whereas ADS-engaged rear-end crashes are more likely to cause minor injuries. However, a notable similarity is that high-speed roads (with posted speed limits of 71 mph or higher) frequently feature animals as crash partners in both ADAS- and ADS-engaged crashes. Based on these findings, it is strongly recommended to focus on infrastructural improvements alongside enhancing AV algorithms and sensor performance, particularly for non-motorist and animal detection in low-light conditions. Policymakers should prioritize driver education on safe AV operation and interaction while also mandating the installation of external human–machine interfaces to enhance AV communication with other road users and reduce rear-end crashes.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"218 ","pages":"Article 108072"},"PeriodicalIF":5.7,"publicationDate":"2025-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143913153","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Exploring the impact of built environment on crash risks at transportation hubs","authors":"Chuanyao Li,&nbsp;Li Chen","doi":"10.1016/j.aap.2025.108079","DOIUrl":"10.1016/j.aap.2025.108079","url":null,"abstract":"<div><div>This study investigates the impact mechanism of the built environment surrounding transportation hubs on crash risks (CR). Three buffer zones (300 m, 500 m, and 800 m) are defined as the spatial analysis units, and Ordinary Least Squares (OLS), Geographically Weighted Regression (GWR), and Multiscale Geographically Weighted Regression (MGWR) are utilized in this study. The results reveals that the 800 m buffer zone provides deeper insights into the factors affecting CR related to the built environment surrounding transportation hubs. Additionally, MGWR demonstrates superior performance in explaining the built environment’s impact on CR compared to the other two methods, with an explanation rate of 83.7 %. To reduce CR near transportation hubs, rationally planning the surrounding land use layout and reducing population density per unit area are recommended. Moreover, the density of road networks surrounding airports and railway stations should be kept at a lower level to reduce CR. The findings of this study contribute to a deeper understanding of the relationship between the built environment surrounding transportation hubs and crashes, providing planning guidance and creating a friendly environment surrounding transportation hubs.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"218 ","pages":"Article 108079"},"PeriodicalIF":5.7,"publicationDate":"2025-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143906425","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Could vehicles analyze driving risks using human fuzzy semantic logic? A data-knowledge-driven new perspective","authors":"Jiming Xie ,&nbsp;Yaqin Qin ,&nbsp;Yan Zhang ,&nbsp;Jianhua Li ,&nbsp;Tianshun Chen ,&nbsp;Xiaohua Zhao ,&nbsp;Yulan Xia","doi":"10.1016/j.aap.2025.108037","DOIUrl":"10.1016/j.aap.2025.108037","url":null,"abstract":"<div><div>Accurate risk identification is crucial for ensuring the safe operation of Host vehicles (HoVs) in environments shared with Neighboring vehicles (NeVs). Traditional risk identification mechanisms typically rely on large amounts of precise numerical data, making it difficult to comprehensively and accurately identify traffic crash risks under conditions of imperfect data associated with fuzzy information. However, human drivers rely on knowledge-driven, subjective assessments using fuzzy descriptors like distance and speed semantics to evaluate driving risk. These insights provide significant value for addressing the limitations of precise data-driven methods. This study proposes a novel traffic crash risk analysis framework called Token Tree Generation and Parsing (TTGP). It integrates knowledge-driven insights from human drivers with data-driven methods. TTGP includes the Token Tree Generation Module (Module 1) and the Token Tree Parsing Module (Module 2). In Module 1, we apply the token-tree-of-thoughts method to transform natural language traffic regulations and vehicles’ traffic behaviors and attribute parameters into token tree based on semantic rules. This module simulates the generation of human fuzzy semantics in traffic scenarios. In Module 2, we integrate three encoders and decoders to extract traffic crash risk semantic features and identify traffic crash risk level from the digitized token tree. Experiments in the highway and urban expressway interweaving areas demonstrate that TTGP can accurately analyze risk using imprecise data. The TTGP performs better than traditional methods such as Tree, Naïve Bayes, RUSBoost and Efficient Logistic Regression models. This study significantly enhances the flexibility, generalization, and reliability of risk assessment. It bridges the gap in how HoVs handle fuzzy information in risk analysis.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"218 ","pages":"Article 108037"},"PeriodicalIF":5.7,"publicationDate":"2025-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143906426","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Collision risk prediction and takeover requirements assessment based on radar-video integrated sensors data: A system framework based on LLM","authors":"Qingchao Liu ,&nbsp;Ruohan Yu ,&nbsp;Yingfeng Cai ,&nbsp;Quan Yuan ,&nbsp;Henglai Wei ,&nbsp;Chen Lv","doi":"10.1016/j.aap.2025.108041","DOIUrl":"10.1016/j.aap.2025.108041","url":null,"abstract":"<div><div>There are safety risks when drivers take over the control of autonomous driving vehicles, and reducing unnecessary takeovers is essential to improve driving safety. This study seeks to develop an interpretable system framework for collision risk prediction and takeover requirements analysis (CPTR-LLM) utilizing a large language model (LLM). The model’s inference performance is enhanced through the collection of extensive perception data and the design of a two-stage training strategy, reasoning chain framework, and an error detection and correction mechanism. In terms of collision risk prediction, the experimental results show that the accuracy of CPTR-LLM can reach 0.88. The Cross-sectional-autoregressive-distributed lag (ARDL) model and Augmented Mean Groups (AMG) confirm the reliability of the model’s predictive performance by revealing the association between different variables and collision risk. Regarding takeover requirement analysis, CPTR-LLM accurately comprehends the characteristics of the pre-takeover scene and comprehensively assesses the takeover requirement level in conjunction with collision risk, thereby effectively reducing unnecessary takeovers in simple driving scenarios and unsafe takeovers in scenarios with multiple moving targets. Overall, the findings of this paper offer significant insights into the application and takeover requirements of LLM in the domain of road safety.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"218 ","pages":"Article 108041"},"PeriodicalIF":5.7,"publicationDate":"2025-05-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143903565","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}