Driving Risk Assessment Research Articles

A spatiotemporal learning approach to safety‐oriented individualized driving risk assessment in a vehicle‐to‐everything (V2X) environment

A spatiotemporal learning approach to safety‐oriented individualized driving risk assessment in a vehicle‐to‐everything (V2X) environment

Prediction of lower limb discomfort of elderly drivers based on key dimensions in the leg space

Significant difference exists in spatial fitness and perception between elderly and non-elderly drivers. However, due to dynamic and real-time changes in human subjective feelings and joint movements, two-dimensional human body templates and human-machine simulation software are not enough to obtain necessary space parameters. In this study, seven key dimensions of the legroom were measured thrice and averaged, in the situation that the seat and posture are comfortable. Such anthropometric data can reflect dynamic perception that may change due to personal emotions and environmental influences. Extreme learning machine (ELM) was adopted to build a prediction model of leg space discomfort degree, and the influence of the activation function and the number of hidden layer neurons on the prediction accuracy of the model were analyzed. In addition, a multiple linear regression (MLR) model was established with the discomfort score as the dependent variable and the seven key dimensions as the independent variables. The results indicated that the ELM model could effectively predict elderly drivers’ discomfort degree (MSE = 0.182, MRE = 9.364, R2 = 0.869) by learning the dimensions of the seven key positions. The MLR model (R2 = 0.861) did not perform as well as ELM. However, the regression coefficients could reflect the degree to which each dimension affects the discomfort degree of leg space for elderly drivers. The conclusions could function in elderly-oriented in-vehicle space arrangement and driving risk assessment of elderly people.

Risk prediction model for distracted driving: Characterizing interactions of eye glances and manual sequences

Distracted driving substantially undermines road safety, and the frequent, low-accuracy distracted driving risk prediction alarms may detrimentally affect drivers’ judgments. The study aims to improve risk prediction accuracy by mapping the distinctive interactions of eye glances and manual sequences in distracted driving into a graph structure of nodes and edges. Distraction patterns are categorized as visual distractions (VD), manual distractions (MD), and visual-manual distractions (VMD) based on daily driving behaviours, such as eating, drinking, reaching for something, and using mobile phones. 1,806 distraction alarm records came from an active safety platform (ASP) in Beijing, covering 69 drivers from 23 hazardous materials road transport companies. The study extracts characteristics to assess distracted driving risks, including visual, manual, and driving performance features. Subsequently, unsupervised learning is used to cluster risk features into three categories (low, medium, and high), which serve as labels for the risk prediction model. In addition, each distraction alarm sequence is divided into nodes and edges of a graph. More specifically, five visual areas, forward (F), object of distraction (Dis), left window (LW), rear-view mirror (RM), center dashboard (C), as well as manual sequences, single hand (H), double hands (2H) are represented as nodes. The edges are connected in parallel (occurring simultaneously) or in series (occurring sequentially), with arrows pointing from the earlier node to the later one. Furthermore, coupled with time, global, and environmental features, a temporal graph attention network (TGAT) with integrated time functions and multi-head attention mechanisms is developed to predict distracted driving risks. The results indicated that VMD demanded more visual and manual resources and led to more high-risk alarms than VD and MD. Besides, TGAT reached a promising result, outperforming other time series methods. This study is valuable for driver distraction monitoring and driving risk assessment, significantly contributing to the enhancement of road safety.

A Driving Risk Assessment Framework Considering Driver’s Fatigue State and Distraction Behavior

A Driving Risk Assessment Framework Considering Driver’s Fatigue State and Distraction Behavior

Autonomous Driving Risk Assessment With Boundary-Based Environment Model

Risk assessment is important for intelligent vehicles to make safe driving decisions. In some researches, the risk is modeled as the effect of each environment element on the ego vehicle. However, the overall understanding of the whole scenario is not achieved. In order to get a comprehensive understanding of the overall scenario, some researchers analyzed risk with 2D field. However, the analysis of surface-level dense risk distribution has heavy computational burden. In view of this, we proposed the boundary-based risk assessment method and corresponding decision-making mechanism, as a light-weighted way of understanding the overall risk distribution for intelligent vehicles. First, different environment elements are generally modeled as the unified environment boundary representing space occupation, semantic type and motion status. Then, the risk in different directions is analyzed according to the environment boundary in the ego-centered polar coordinate. Furthermore, considering the perception uncertainty, the safe boundary is set to serve for decision-making as boundary constraint. Experiment results proved that the proposed boundary-based risk assessment method has significantly reduced computational burden than surface-level dense risk assessment with similar decision performance.

A Review of Intelligent Systems for Driving Risk Assessment

A Review of Intelligent Systems for Driving Risk Assessment

Review of Intelligent Vehicle Driving Risk Assessment in Multi-Vehicle Interaction Scenarios

With the rapid breakthroughs in artificial intelligence technology and intelligent manufacturing technology, automotive intelligence has become a research hotspot, and much progress has been made. However, a skeptical attitude is still held towards intelligent vehicles, especially when driving in a complex multi-vehicle interaction environment. The interaction among multi-vehicles generally involves more uncertainties in vehicle motion and entails higher driving risk, and thus deserves more research concerns and efforts. Targeting the safety assessment issue of complex multi-vehicle interaction scenarios, this article summarizes the existing literature on the relevant data collection methodologies, vehicle interaction mechanisms, and driving risk evaluation methods for intelligent vehicles. The limitations of the existing assessment methods and the prospects for their future development are analyzed. The results of this article can provide a reference for intelligent vehicles in terms of timely and accurate driving risk assessment in real-world multi-vehicle scenarios and help improve the safe driving technologies of intelligent vehicles.

A Dynamic Path-Planning Method for Obstacle Avoidance Based on the Driving Safety Field.

Establishing an accurate and computationally efficient model for driving risk assessment, considering the influence of vehicle motion state and kinematic characteristics on path planning, is crucial for generating safe, comfortable, and easily trackable obstacle avoidance paths. To address this topic, this paper proposes a novel dual-layered dynamic path-planning method for obstacle avoidance based on the driving safety field (DSF). The contributions of the proposed approach lie in its ability to address the challenges of accurately modeling driving risk, efficient path smoothing and adaptability to vehicle kinematic characteristics, and providing collision-free, curvature-continuous, and adaptable obstacle avoidance paths. In the upper layer, a comprehensive driving safety field is constructed, composed of a potential field generated by static obstacles, a kinetic field generated by dynamic obstacles, a potential field generated by lane boundaries, and a driving field generated by the target position. By analyzing the virtual field forces exerted on the ego vehicle within the comprehensive driving safety field, the resultant force direction is utilized as guidance for the vehicle's forward motion. This generates an initial obstacle avoidance path that satisfies the vehicle's kinematic and dynamic constraints. In the lower layer, the problem of path smoothing is transformed into a standard quadratic programming (QP) form. By optimizing discrete waypoints and fitting polynomial curves, a curvature-continuous and smooth path is obtained. Simulation results demonstrate that our proposed path-planning algorithm outperforms the method based on the improved artificial potential field (APF). It not only generates collision-free and curvature-continuous paths but also significantly reduces parameters such as path curvature (reduced by 62.29% to 87.32%), curvature variation rate, and heading angle (reduced by 34.11% to 72.06%). Furthermore, our algorithm dynamically adjusts the starting position of the obstacle avoidance maneuver based on the vehicle's motion state. As the relative velocity between the ego vehicle and the obstacle vehicle increases, the starting position of the obstacle avoidance path is adjusted accordingly, enabling the proactive avoidance of stationary or moving single and multiple obstacles. The proposed method satisfies the requirements of obstacle avoidance safety, comfort, and stability for intelligent vehicles in complex environments.

A generalized driving risk assessment on high-speed highways using field theory

This study presents a new safety measure derived from field theory. It evaluates the risk arising from the various concurrent conflicts within a platoon that can occur on high-speed highway driving situations, such as car-following, yielding, and lane changing. We defined the risk field as a finite scalar field produced by traveling vehicles on the road, and we defined the conflict field as the overlapping risk field between any vehicles in proximity on the roadway. The study used a probabilistic trajectory prediction model to construct risk fields and an approximation method to reduce the computational time for real-time applications. To demonstrate the applicability of the proposed new measure, we applied it to real-world trajectory data (NGSIM data from US Highway 101). We compared the results with three traditional conflict-based safety measures: post-encroachment time (PET), modified time-to-collision (MTTC), and deceleration rate to avoid a crash (DRAC). The new measure produced seamless and continuous risk estimations even during time windows when the other measures could not estimate the risk between vehicles. This is a major advantage over traditional measures. The study also developed visual displays of the estimated conflict fields to provide safety analysts with an intuitive and fast understanding of the results of the safety assessments made using the conflict field measure. We conclude that the proposed new safety measure provides a robust, reliable, and improved assessment of the risk involved in expected future mixed-traffic environments that involve both human-driven vehicles and automated vehicles in the future.

Review of Computational Techniques for Modelling Eco-Safe Driving Behavior

Driving is a complex task involving the perception of the driving event, planning response, and action. Safe driving ensures the vehicle’s and its passengers’ safety, whereas economical driving brings down fuel consumption. Eventually, eco-safe driving that ensures economical as well as safe driving is the best bet. This review paper provides a systematic comprehensive analysis across cross-cutting dimensions such as data collection mechanisms, features affecting eco-safe driving, computational models for driving behavior analysis, driver motivational approaches towards eco-safe driving, exploiting research gaps and opportunities for further research in this domain. Driving behavior along with environmental context, including weather information, road conditions, traffic flow and time of travel, represent the most effective factors for doing eco-safe driving analysis. 82% of reviewed papers recommended OBD as a reliable data collection source, along with supplementary information from body sensors and cameras. The K-Mean clustering is an effective driving profiling technique clubbed with dimensionality reduction techniques based on Random Forest regressor, PCA and autoencoders. Deep learning and ensemble learning-based safety approaches utilizing Recurrent Convolutional Networks (RCN), Convolutional Neural Networks (CNN), and Long Short-Term Memory (LSTM) and Decision Tree (DT) have achieved impressive accuracies surpassing 99%, followed by Neural Networks (NN), Support Vector Machines (SVM) and Random Forest (RF) with accuracy ranging from 91% to 96%. Long Short-Term Memory (LSTM) yielded superior Area Under Curve (AUC of 0.836) for fuel prediction, in comparison to Support Vector Machines (SVM) and Neural Networks (NN). Gated Recurrent Unit (GRU) represents fine-grained accurate fuel-prediction methods with accuracy comparable to Long Short-Term Memory (LSTM). Prominent research gaps identified during this study are the lack of a comprehensive approach covering all aspects related to safety, fuel economy, the scope of improvement in real-time driving risk assessment at appropriate granularity level, missing effective and engaging driving feedback, dealing with uncertain and incomplete driving events, driver’s personal traits affecting driving safety and fuel-economy. The review will help in establishing the readiness of automation of driving analysis for reinforcement of eco-safe driving for various kinds of vehicles plug-in hybrid vehicles, hybrid electric vehicles, electric vehicles, and self-driving cars.

Real-time driving risk assessment based on the psycho-physical field

Real-time Driving Risk Assessment (RDRA) is a critical component of traffic safety and is influenced by confounding impacts from drivers, surrounding vehicles, and roadway conditions. Previous studies simplified the RDRA based on kinematic characteristics. However, drivers perceived the risks not only through kinematic characteristics but also by anticipating the behaviors of surrounding participants and execute collision-preventive maneuvers. In this study, an innovative RDRA framework based on the Psycho-Physical Field (PPF) is proposed. Specifically, the PPF of the subjective vehicle exerts repulsive forces on intrusive risk sources in different directions, of which the two-dimensional field distribution is determined by physical collision-related measures and regulated by the psychological characteristics of behavior anticipations and risk perception abilities. The proposed method was first analyzed through theoretical feasibility analyses and verified for over 450 high-risk events from naturalistic driving data, including three typical types of scenarios: car-following, lane-changing, and being cut-in/off. Moreover, the adaptability was further validated through three cases and compared with the risk warning functions of Mobileye. The results showed that the proposed method can provide accurate risk evaluations, and identify potential hazards about 2 s in advance for high-risk cut-in events.

Driving Decisions for Autonomous Vehicles in Intersection Environments: Deep Reinforcement Learning Approaches with Risk Assessment

Intersection scenarios are one of the most complex and high-risk traffic scenarios. Therefore, it is important to propose a vehicle driving decision algorithm for intersection scenarios. Most of the related studies have focused on considering explicit collision risks while lacking consideration for potential driving risks. Therefore, this study proposes a deep-reinforcement-learning-based driving decision algorithm to address these problems. In this study, a non-deterministic vehicle driving risk assessment method is proposed for intersection scenarios and introduced into a learning-based intelligent driving decision algorithm. In addition, this study proposes an attention network based on state information. In this study, a typical intersection scenario was constructed using simulation software, and experiments were conducted. The experimental results show that the algorithm proposed in this paper can effectively derive a driving strategy with both driving efficiency and driving safety in the intersection driving scenario. It is also demonstrated that the attentional neural network designed in this study helps intelligent vehicles to perceive the surrounding environment more accurately, improves the performance of intelligent vehicles, as well as accelerates the convergence speed.

A Study on the Development of Driving Risk Assessment Model for Autonomous Vehicles Using Fuzzy-AHP

A Study on the Development of Driving Risk Assessment Model for Autonomous Vehicles Using Fuzzy-AHP

Using contextual data to predict risky driving events: A novel methodology from explainable artificial intelligence

Usage-based insurance has allowed insurers to dynamically tailor insurance premiums by understanding when and how safe policyholders drive. However, telematics information can also be used to understand the driving contexts experienced by the driver within each trip (e.g., road types, weather, traffic). Since different combinations of these conditions affect exposure to accidents, this understanding introduces predictive opportunities in driving risk assessment. This paper investigates the relationships between driving context combinations and risk using a naturalistic driving dataset of 77,859 km. In particular, XGBoost and Random Forests are used to determine the predictive significance of driving contexts for near-misses, speeding and distraction events. Moreover, the most important contextual factors in predicting these risky events are identified and ranked through Shapley Additive Explanations. The results show that the driving context has significant power in predicting driving risk. Speed limit, weather temperature, wind speed, traffic conditions and road slope appear in the top ten most relevant features for most risky events. Analysing contextual feature variations and their influence on risky events showed that low-speed limits increase the predicted frequency of speeding and phone unlocking events, whereas high-speed limits decrease harsh accelerations. Low temperatures decrease the expected frequency of harsh manoeuvres, and precipitations increase harsh acceleration, harsh braking, and distraction events. Furthermore, road slope, intersections and pavement quality are the most critical factors among road layout attributes. The methodology presented in this study aims to support road safety stakeholders and insurers by providing insights to study the contextual risk factors that influence road accident frequency and driving risk.

Decline in Other Instrumental Activities of Daily Living as Indicators of Driving Risk in Older Adults at an Academic Memory Clinic.

Decisions around driving retirement are difficult for older persons living with cognitive decline and their caregivers. In many jurisdictions, physicians are responsible for notifying authorities of driving risks. However, there are no standardized guidelines for this assessment. Having access to a driving risk assessment tool could help older adults and their caregivers prepare for discussions around driving retirement. This study compares the clinical profiles of older adult drivers assessed in an academic memory clinic who were referred to the driving authority to older drivers who were not with a focus on instrumental activities of daily living (iADLs). Data on referred (R) and not-referred (NR) drivers were extracted from medical records. Elements from the medical history, cognitive history, functional abilities, Modified Mini-Mental State (3MS) examination, Trails A/B, and clock drawing were included in the analysis. Four risk factors of interest were examined in separate logistic regression analyses, adjusted for demographic variables. 50 participants were identified in each group. The R group was older on average than the NR. As expected, R were more likely to have Trails B scores over 3 min and have significantly abnormal clock drawing tests. R also showed lower 3MS scores and a higher average number of functional impairments (including managing appointments, medications, bills, or the television). Beyond standard cognitive tests, impairment in iADLs may help general practitioners identify at-risk drivers in the absence of standardized guidelines and tools. This finding can also inform the design of a risk assessment tool for driving and could help with approaches for drivers with otherwise borderline test results.

Driving risk assessment under the connected vehicle environment: a CNN-LSTM modeling approach

Driving risk assessment under the connected vehicle environment: a CNN-LSTM modeling approach

Relationship Between Driving Risk Assessment Test Using Touch Panel Display and Compensation Intention and Driving Behavior

Relationship Between Driving Risk Assessment Test Using Touch Panel Display and Compensation Intention and Driving Behavior

Real-time detection of abnormal driving behavior based on long short-term memory network and regression residuals

Abnormal driving behavior is one of the main causes of roadway collisions. In most studies of abnormal driving behavior, the abnormal driving status is detected and analyzed using classification algorithms directly or using unsupervised learning algorithms to classify reconstruction or prediction residuals. However, abnormal driving behavior data are difficult to acquire and label. Also, a class imbalance issue is inherent in the algorithm training process due to the relatively sparse data for abnormal driving behavior. Moreover, current studies that include residual analysis tend to focus on individual points and thus fail to capture the continuity characteristic of abnormal driving behavior.To address these problems, a long short-term memory-residual (LSTM-R) algorithm is proposed to detect abnormal driving behavior in real time. The proposed algorithm (referred to simply as LSTM-R) has two steps. First, an LSTM network is used to fit the current vehicle kinematic data based on historical data to obtain the root mean square residual at each moment. Second, a time window-based residual algorithm is designed and employed to detect abnormal driving behavior according to the magnitude and continuity of the residuals. To verify the effectiveness of LSTM-R, an experimental test was conducted in Nanjing, China. The vehicle kinematic data were collected non-intrusively using a smartphone.In addition, AdaCost, SMOTEBoost, EasyEnsemble, LightGBM-residual, and linear regression-residual algorithms were employed for comparison with the proposed algorithm to assess its effectiveness. The effects of (1) the degree-of-fit of the LSTM network, (2) the LSTM-R parameters, and (3) the abnormal driving behavior percentage on the detection results were analyzed in detail. First, both the underfitting and overfitting of the LSTM network compromise the detection performance. Second, within a certain range of values, the LSTM-R parameters have little effect on the detection results. Third, the detection results are affected only slightly by the abnormal proportion. The results show that LSTM-R, with a maximum F1-score of 0.866, significantly outperforms the other five algorithms. Furthermore, even if only 10% abnormal driving behavior is in the training set, LSTM-R’s F1-score can still be close to 0.86, indicating a significant relaxation of the requirements for labeled data. Also, the required data are easy to collect, which indicates LSTM-R’s extensive application possibilities. This paper thus provides an effective method for the real-time detection of abnormal driving behavior and also supports driving risk assessment and driving behavior improvement with the overall goal to enhance roadway safety.

Driving Risk Assessment Using Non-Negative Matrix Factorization With Driving Behavior Records

Aggressive driving behavior (ADB) is a major cause of traffic accidents. As ADB is controllable, ADB-based driving risk assessment is an effective method for drivers and transportation companies to ensure driving safety. Conventionally, the relationships between ADBs and accident-related records are analyzed when assessing driving risk. However, such records typically overlook driver responsibility for driving risks and depend considerably on the person producing the data (e.g., police officers or insurance managers). Foremost, conventional approaches do not consider non-accident situations that comprise most driving scenarios. Thus, we propose a novel driving risk assessment method that uses only ADB data. In this method, interpretable latent risk factors are extracted from ADB data via sparse non-negative matrix factorization (NMF), and then the driving risk score is computed on a scale of 0–100. The proposed method was validated by adopting a real-world application to assess the driving risk of bus drivers in South Korea and by conducting an evaluation performed by transportation experts in conjunction with the Korea Transportation Safety Authority. Results revealed that the proposed method can discriminate between high- and low-risk driving, thus providing clear guidelines to improve driving. Then, the proposed driving risk score assessment method using NMF was compared with existing machine learning-based risk assessment methods. The proposed method outperformed the conventional methods in terms of driving risk discrimination and interpretability. This study can provide risk assessment guidelines based on driving behavior records and contribute to the application of machine learning in transportation safety management.

Environmentally driven risk assessment for algal bloom occurrence in shallow lakes

An algal bloom is a complex hydro-biological phenomenon driven by multi-attribute environmental processes and thus is still difficult to predict. In this paper, a comprehensive modelling framework for forecasting algal bloom risks in shallow lakes is presented, which is based on long-term field observation and modelling of eutrophic shallow lakes. In the procedure, the major factors and their suitable ranges are investigated, and the individual influence of various driving factors is evaluated quantitatively, using an integrated approach of orthogonal design and regression analysis. By analysing the possible combined effects of the major driving factors and the relationship between algal bloom risk and major bloom-driving factors, a cost-effective environmentally driven risk assessment model is developed to forecast the likelihood of algal bloom occurrence, through a parameter optimization and prediction comparison routine. The risk model has been calibrated and validated against long-term field observations of algal blooms in Taihu Lake, with the prediction accuracy higher than 70%, which only requires readily available meteorological and water quality data. It is noted that for the closed shallow lake, the influence of hydrodynamics can be indirectly reflected by the variation of wind speed; and, total phosphorus, water temperature, photosynthetically active radiation, and average wind speed could be used as major bloom-driving factors in Taihu Lake generally. This study provides a practical framework for the development of algal bloom early warning schemes for shallow lakes and helps to understand the combined function of complex bloom-driving factors.