Accident; analysis and prevention最新文献

{"title":"Learning salient representation of crashes and near-crashes using supervised contrastive variational autoencoder","authors":"Boyu Jiang ,&nbsp;Feng Guo","doi":"10.1016/j.aap.2025.108148","DOIUrl":"10.1016/j.aap.2025.108148","url":null,"abstract":"<div><div>Models capable of learning representations that are salient in safety–critical events (SCEs; including crashes and near-crashes) are crucial for road safety. This study proposes a novel deep learning model, the supervised contrastive variational autoencoder (scVAE), that incorporates supervised contrastive learning methods into the variational autoencoder (VAE) framework. By leveraging two distinct encoders, the scVAE encourages the salient latent variables to be discriminative, capturing the unique representations of SCEs while being regulated by the response variable to focus on the most relevant representations for accurate clustering. Through application on the Second Strategic Highway Research Program (SHRP 2) Naturalistic Driving Study kinematic datasets, we demonstrated the effectiveness of the scVAE in learning salient representations that enable improved clustering compared to alternative models. Quantitative analysis revealed clear clustering patterns in the learned salient representation space, facilitating downstream tasks such as generating samples, denoising, and prediction. The proposed approach of combining contrastive and supervised learning can be scaled to other model frameworks and data modalities, offering a promising direction for learning-enhanced representations that cater to tasks of interest. The study findings highlight the contributions of scVAE to traffic safety, offering enhanced capabilities for driving scenario generation and SCE detection.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"220 ","pages":"Article 108148"},"PeriodicalIF":5.7,"publicationDate":"2025-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144517426","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":"Safety and efficiency-oriented adaptive strategy controls for connected and automated vehicles in unstable communication environment","authors":"Yangzhen Zhao ,&nbsp;Xuedong Hua ,&nbsp;Weijie Yu ,&nbsp;Wenxie Lin ,&nbsp;Wei Wang ,&nbsp;Qihao Zhou","doi":"10.1016/j.aap.2025.108121","DOIUrl":"10.1016/j.aap.2025.108121","url":null,"abstract":"<div><div>Connected and automated vehicle (CAV) has demonstrated significant potential in enhancing traffic safety, efficiency, and environmental sustainability in the premise of high-quality communication environment. However, attributed to unexpected disturbances in real life, vehicular communication often suffers from unstable information transmission that lowers the performance of CAV platooning. Despite this drawback, existing research lacks sufficient attention on CAV platooning in the presence of unstable communication environment (UCE) and rarely addresses the response strategies. To fill these research gaps, this study integrates dynamic communication topologies into multi-class controls of automated and/or connected vehicles in response to various data abnormalities, based on simulation data generated from the modified intelligent driver model (IDM) and the communication disturbance framework designed to emulate UCE. The impact of UCE on CAV platooning performance is evaluated in terms of traffic safety and efficiency. The study introduces five adaptive dynamic strategy controls (ADSCs) and three hybrid switching ADSCs to enhance platoon performance under varying UCE conditions, along with recommendations on applicable ADSCs for maintaining the platoon performance under different UCE scenarios. Also, our sensitivity analysis of attenuation model parameters identifies optimal configurations for the development of ADSCs. Results show that continuous UCE (CUCE) poses greater safety challenges compared to intermittent UCE (IUCE), with combined IUCE being the most detrimental. Scenarios involving aggressive acceleration behavior significantly increase the collision risk and speed oscillation within the platoon. The proposed hybrid switching ADSCs significantly improve CAV platoon performance, offering resilient and adaptive vehicle control strategies to smooth traffic flow under disturbances.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"220 ","pages":"Article 108121"},"PeriodicalIF":5.7,"publicationDate":"2025-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144517601","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":"Directional vibrotactile takeover requests on a wrist-worn device: effects of age, pattern type, and urgency in automated driving","authors":"Wei-Hsiang Lo,&nbsp;Gaojian Huang","doi":"10.1016/j.aap.2025.108093","DOIUrl":"10.1016/j.aap.2025.108093","url":null,"abstract":"<div><div>Drivers are still required to perform the takeover task in highly automated vehicles. This task, which is cognitively and physically demanding, may present challenges for older adults due to general age-related declines in perception and cognition. Tactile modalities that may not be occupied by many non-driving-related tasks could serve as a potential solution for delivering takeover requests. Among these, directional vibrotactile stimuli presented via a wrist-worn device represent a promising approach. However, the effects of the two common types of directional vibrotactile patterns, dynamic patterns that vibrate sequentially at different locations and static patterns that vibrate at fixed locations, are still unknown. Therefore, this study aimed to investigate the effect of age (younger and older adults), vibrotactile pattern types (Baseline, Full-Dynamic, Semi-Dynamic, and Static), and interpulse interval (shorter (300 ms) and longer (800 ms)) on takeover performance. Forty participants (20 younger and 20 older adults) were engaged in the SAE Level 3 driving simulator study. Overall, Static and Baseline patterns were associated with faster reaction and takeover times and were perceived as more useful and satisfactory compared to the Full-Dynamic and Semi-Dynamic patterns. Shorter interpulse intervals (300 ms) for vibrotactile takeover requests resulted in better takeover performance, as indicated by shorter reaction and takeover times compared to longer interpulse intervals (800 ms). Finally, younger adults reacted faster to vibrotactile takeover requests than older adults did. The findings from the current study may inform the design of human–machine interfaces on wearable devices for next-generation automated vehicles.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"220 ","pages":"Article 108093"},"PeriodicalIF":5.7,"publicationDate":"2025-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144502367","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":"Clustering of freeway cut-in scenarios for automated vehicle development considering data dimensionality and imbalance","authors":"Penghui Li ,&nbsp;Qianru Dong ,&nbsp;Xiangjun Zhao ,&nbsp;Chao Lu ,&nbsp;Mengxia Hu ,&nbsp;Xuedong Yan ,&nbsp;Chunjiao Dong","doi":"10.1016/j.aap.2025.108151","DOIUrl":"10.1016/j.aap.2025.108151","url":null,"abstract":"<div><div>Representative driving scenarios derived by clustering of naturalistic driving data are guidelines for the function definition and algorithm development of automated vehicle. However, current clustering methods struggle with data dimensionality and imbalance, leading to significant biases. To tackle these issues, this study proposed a novel two-layer self-adaptive multiprototype-based competitive learning algorithm, and implemented it in clustering of freeway cut-in scenarios. Firstly, the extracted cut-in segments from naturalistic driving data included environmental, static, and dynamic vehicle elements, composed of discrete, continuous, and time series variables, posing a challenge in multi-dimensional parameter clustering. To tackle this, we utilized the K-medoids clustering method, based on dynamic time warping distance, to cluster variables such as cut-in vehicle velocity, converting them into discrete variables and applying one-hot encoding for easier clustering distance calculations. Secondly, to address the imbalance issue where minority sample categories were absorbed into majority types in naturalistic driving data clustering, we employed a multi-prototype clustering method in the second layer. Each cluster was represented by one or more sub-clusters to ensure adequate representation of minority clusters. Moreover, the inclusion of adaptive competitive learning allowed the algorithm to autonomously determine the optimal number of clusters, eliminating the need for manual parameter tuning. Consequently, the proposed algorithm produced eleven representative freeway cut-in scenarios from 2415 segments, with a better clustering goodness than the other traditional clustering methods. Moreover, four representative cut-in scenarios were frequently appeared in the dataset and commonly recognized by previous studies, whilst seven were rare in the dataset but common in real-world driving circumstances, such as at night, adverse weather conditions, and commercial vehicle cut-in scenarios. These findings suggest that the proposed clustering method effectively addresses the challenges of dimensionality and imbalance, indicating its potential for wide application in constructing representative scenarios for automated vehicles development.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"220 ","pages":"Article 108151"},"PeriodicalIF":5.7,"publicationDate":"2025-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144502368","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":"An interpretable stacking ensemble learning model for visual-manual distraction level classification for in-vehicle interactions","authors":"Yahui Wang ,&nbsp;Zhoushuo Liang ,&nbsp;Pengfei Tian ,&nbsp;Yue He","doi":"10.1016/j.aap.2025.108150","DOIUrl":"10.1016/j.aap.2025.108150","url":null,"abstract":"<div><div>Recognizing the level of driver distraction during the execution of secondary tasks within the intelligent cockpit is crucial for ensuring a seamless interaction between human drivers and intelligent vehicle systems. To address this issue, this paper proposes a framework for recognizing driver distraction levels that integrates clustering, classification, and interpretability. First, Feature Selection with Optimal Graph(SF²SOG) is employed to identify discriminative features from the data facilitating dimensionality reduction. Following this, Agglomerative Clustering are employed to classify the gathered unlabeled data on driving distraction behaviors into three distinct categories. Additionally, a heuristic-based stacking ensemble model is introduced to identify the levels of driver distraction, using the factors that influence these levels as input parameters for classification. To improve the effectiveness of the stacking model, three diverse classifiers adaptive boosting (AdaBoost)、random forest (RF) and extreme gradient boosting (XGBoost) are selected to serve as the base models, while a relatively simple yet accurate model logistic regression (LR) is used as the <em>meta</em>-classifier. Finally, Shapley Additive exPlanations (SHAP) is employed for interpretability analysis. Notably, heuristic-based stacking ensemble model achieved a commendable accuracy rate of 96.25%, highlighting its significant advantage. Further analysis shows that higher maximum pupil diameter (maxPD) and mean pupil diameter (meanPD) indicate increased distraction, while greater glance frequency and lane deviation reflect reduced situational awareness and control. These findings are crucial for reducing accidents and enhancing driving safety in the era of intelligent vehicles.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"220 ","pages":"Article 108150"},"PeriodicalIF":5.7,"publicationDate":"2025-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144489568","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":"Learnable Operational Design Condition monitor for failure prediction in autonomous driving","authors":"Haowei Xu ,&nbsp;Yougang Bian ,&nbsp;Yang Li ,&nbsp;Hongmao Qin ,&nbsp;Hanchu Zhou ,&nbsp;Fangrong Chang ,&nbsp;Shaofei Wang ,&nbsp;Qing Ye","doi":"10.1016/j.aap.2025.108139","DOIUrl":"10.1016/j.aap.2025.108139","url":null,"abstract":"<div><div>Autonomous vehicle accidents frequently originate from violations of Operational Design Conditions (ODC)—the predefined operational limits of vehicle states, environmental factors, and driver capabilities. ODC violation indicates a high probability of impending functional failure under current operational scenarios, where failure denotes the system’s inability to achieve designated performance thresholds or maintain safety-critical constraints. Therefore, designing and continuously monitoring the boundary states of ODC during system operation constitutes a critical imperative to prevent system failures, ensure operational safety, and mitigate autonomous vehicle accidents. However, prevailing ODC monitoring methods primarily rely on first-order logic checklists, failing to capture emergent risks from parameter interactions. Therefore, this paper establishes an end-to-end methodological framework for constructing a learnable ODC monitor termed ODCNet to realize failure prediction. The architecture first projects operational states into unified latent representations, then derives probabilistic boundary estimates through neural inference, and finally calibrates residual errors via hybrid Gaussian Process regression. Additionally, an adaptive active learning mechanism continuously refines boundary precision through targeted testing of high-uncertainty scenarios. The validation through intersection, lane-keeping, and vehicle detection case studies demonstrates the failure prediction performance of the ODC monitor that precedes accidents.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"220 ","pages":"Article 108139"},"PeriodicalIF":5.7,"publicationDate":"2025-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144489632","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 sensory distractions on reaction time: a comprehensive modeling approach","authors":"Rajesh Chouhan ,&nbsp;Yogesh Bhagwan Chavan ,&nbsp;Ashish Dhamaniya ,&nbsp;Constantinos Antoniou","doi":"10.1016/j.aap.2025.108154","DOIUrl":"10.1016/j.aap.2025.108154","url":null,"abstract":"<div><div>Auditory distractions can significantly impact driving safety by affecting a driver’s ability to focus, react, and make decisions, and consequently increase Driver’s Reaction Time. The present study aims to examine the impact of such distractions using experimental data employing the Vienna Test System (VTS). Eighty drivers (70% male, 30% female) from mixed traffic environments participated, resulting in 240 data samples across two age groups (young and mature). Reaction time was assessed under three conditions: (i) Normal, (ii) Music, and (iii) Call using the VTS, which includes visual and auditory stimuli requiring physical reactions, thus closely simulating real-life driving scenarios. Results showed that male drivers’ reaction time increased by 4.8% and 20.1% under Music and Call conditions, respectively, while female drivers exhibited increases of 6.1% and 13.5%, respectively. The comparison of reaction time among young and mature groups under each condition indicated that Call distractions led to the highest increase (12.0%) in reaction time, followed by Normal (10.8%) and Music (8.3%). A second-degree polynomial equation was proposed to estimate reaction time from age. Additionally, the Weibull Accelerated Failure Time (AFT) model identified key factors affecting reaction time under normal conditions, resulting in a formulated mathematical expression. A validation experiment using real-world crash and near-crash videos reinforced the findings. The study provides empirical insights for enhancing safety measures and could be leveraged within advanced driver assistance systems (ADAS).</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"220 ","pages":"Article 108154"},"PeriodicalIF":5.7,"publicationDate":"2025-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144489631","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":"A perspective from competitive-cooperative driving modes: Identification of vehicle merging behavior models and crash risk factors in merge zone","authors":"Nengchao Lyu ,&nbsp;Siqi Feng ,&nbsp;Hongliang Wan","doi":"10.1016/j.aap.2025.108153","DOIUrl":"10.1016/j.aap.2025.108153","url":null,"abstract":"<div><div>In highway and expressway merge zones, merge behavior is the most fundamental maneuvering behavior and is a major cause of traffic conflicts and collisions. While some studies have employed vehicle trajectory data to simulate and investigate microscopic merge behavior, most have failed to consider the interaction between vehicles and have not fully investigated the factors that influence merge risk using vehicle trajectory data. This study considers the interactions between vehicles during the merging process and, based on a “competitive-cooperative” driving mode perspective, proposes a new method for classifying vehicle merging behavior modes in merging zones. It also explores the differences in risk levels and influencing factors between different merging modes. Using aerial-captured merging vehicle trajectory data, and based on previous research and actual observations, merging behaviors are classified into free, homeostatic, competitive, and cooperative modes (FM, HM, CmM, and CoM) according to the gap distance and microscopic interaction between vehicles before and after merging. The lane change risk index (LCRI) is introduced as an alternative safety measure to assess the risk of merging vehicle groups. The k-means algorithm is used to classify the LCRI, and a separate ordered logit model is constructed for each merging behavior mode. The modeling results indicate that classification modeling performs better than all data modeling, and the risk level of merging vehicle groups is closely related to the motion state of the merge vehicle, the average and standard deviation of the longitudinal speed difference between vehicles, and the gap distance between vehicles. The risk levels of the four merging modes and their influencing factors differ, with FM having the lowest risk, followed by CoM, HM having a higher risk, and CmM having the highest risk. The study results provide an in-depth analysis of merge zone risk factors, offering a theoretical basis for improving traffic safety management measures.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"220 ","pages":"Article 108153"},"PeriodicalIF":5.7,"publicationDate":"2025-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144489629","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":"The Halo effect in airport terminals: how wayfinding experiences influence emergency preparedness through perceived reliability","authors":"Botao Zhang ,&nbsp;Yangzhi Yan ,&nbsp;Wei Xie ,&nbsp;Xiaowei Luo ,&nbsp;Eric Wai Ming Lee ,&nbsp;Xiaopeng Deng","doi":"10.1016/j.aap.2025.108149","DOIUrl":"10.1016/j.aap.2025.108149","url":null,"abstract":"<div><div>As critical infrastructure in global transportation networks, airport terminals with high passenger throughput require effective safety management to ensure passenger safety and operational efficiency. Recognizing the importance of a human-centric approach, this study delves into wayfinding signage (WS) that facilitates routine navigation for passengers in airport terminals and emergency exit signage (EES) that guides evacuations during emergencies. We hypothesize that the passengers’ experience in different WS performance criteria—visibility, legibility, consistency, and correctness—affect passengers’ perceived reliability of it, which, by virtue of the Halo Effect, subsequently influences their perceived reliability of EES and willingness to comply with EES. To validate the hypotheses in our research model, we designed a targeted questionnaire and collected 397 valid responses from airport terminals in four major cities in China. The results substantiate the presence of the Halo Effect and reveal the significant impact of certain dimensions of WS performance criteria on passengers’ reliability perception and their willingness to comply. The findings emphasize the mediating role of perceived reliability in converting environmental design into behavioral intentions and actions, offering theoretical insights and practical recommendations for optimizing WS design to improve routine services and emergency preparedness in airport management. It also advances the understanding of reliability perception mechanisms in public spaces and offers practical significance for improving public safety through everyday efforts.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"220 ","pages":"Article 108149"},"PeriodicalIF":5.7,"publicationDate":"2025-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144489630","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":"How to assess situation awareness while driving with automation? Impact of visual attention during AD mode on measures of situation awareness","authors":"Barbara Metz ,&nbsp;Johanna Wörle ,&nbsp;Myriam Metzulat ,&nbsp;Alexandra Neukum","doi":"10.1016/j.aap.2025.108142","DOIUrl":"10.1016/j.aap.2025.108142","url":null,"abstract":"<div><div>A variety of objective and subjective methods is used to assess situation awareness (SA) in automated driving (AD). In fields like aviation, established methods like SAGAT are used as state of the art for assessing SA, whereas in AD, there is less consensus on the most valid measures. In a driving simulator study with N = 41 participants, four different levels of SA were experimentally created by manipulating visual attention during AD. Different methods for assessing SA were logged during AD-mode and during takeover situations including subjective ratings, gaze behaviour, performance and probe measures. The impact of the manipulation of visual attention during AD mode on the different measures of SA as well as their relation to each other is analysed, reported and discussed. Results show pronounced differences between levels of attention during AD-mode in subjective SA, gaze behaviour and performance measures while preventing visual processing of the driving scenery completely while during AD mode caused surprisingly little impact. A relation between measures of SA and performance can be shown for specifically designed takeover scenarios with increased demands on SA. On the one hand, this implies that there can be a critical impact of SA on performance in takeover situations but it also highlights the robustness and efficiency of drivers’ visual processing that enables safe takeover responses even in situations with only little visual processing before a takeover. The implications of the results for assessing SA as well as the processes behind SA in AD are discussed.</div></div>","PeriodicalId":6926,"journal":{"name":"Accident; analysis and prevention","volume":"220 ","pages":"Article 108142"},"PeriodicalIF":5.7,"publicationDate":"2025-06-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144365354","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}