Historical Crash Research Articles

Investigating surrogate safety measures’ threshold consistency on different types of curves

Surrogate safety measures (SSM) are widely used in safety analysis. In addition to offering the advantages of being a proactive approach, it also addresses the limitations of historical crash data, including inaccuracies and unavailability. Most SSM studies focus particularly on tangents and intersections, but curve-specific analyses remain underexplored. This study aims to investigate whether SSM thresholds remain consistent across all types of curves or whether adjustments are necessary. After analyzing 5600 km of road network in Kerala, India, we observe that the curves in it can be classified into 19 well-defined clusters based on geometric and infrastructural attributes such as radius, length, sight distance, abutting land use, and number of access points. We extract trajectories using aerial videography in four selected curve clusters and use Anticipated Collision Time (ACT) for conflict analysis. We compare traffic conflicts estimated at various ACT values with actual crash counts across all four curve clusters to determine the SSM threshold. Observing differences in conflict-based and historic crash–based vulnerability rankings, we illustrate that different thresholds are warranted in different types of curves. We further present an optimization formulation to estimate these thresholds by comparing the relative vulnerability of curve clusters.

Integrating LiDAR Sensor Data into Microsimulation Model Calibration for Proactive Safety Analysis.

Studies have shown that vehicle trajectory data are effective for calibrating microsimulation models. Light Detection and Ranging (LiDAR) technology offers high-resolution 3D data, allowing for detailed mapping of the surrounding environment, including road geometry, roadside infrastructures, and moving objects such as vehicles, cyclists, and pedestrians. Unlike other traditional methods of trajectory data collection, LiDAR's high-speed data processing, fine angular resolution, high measurement accuracy, and high performance in adverse weather and low-light conditions make it well suited for applications requiring real-time response, such as autonomous vehicles. This research presents a comprehensive framework for integrating LiDAR sensor data into simulation models and their accurate calibration strategies for proactive safety analysis. Vehicle trajectory data were extracted from LiDAR point clouds collected at six urban signalized intersections in Lubbock, Texas, in the USA. Each study intersection was modeled with PTV VISSIM and calibrated to replicate the observed field scenarios. The Directed Brute Force method was used to calibrate two car-following and two lane-change parameters of the Wiedemann 1999 model in VISSIM, resulting in an average accuracy of 92.7%. Rear-end conflicts extracted from the calibrated models combined with a ten-year historical crash dataset were fitted into a Negative Binomial (NB) model to estimate the model's parameters. In all the six intersections, rear-end conflict count is a statistically significant predictor (p-value < 0.05) of observed rear-end crash frequency. The outcome of this study provides a framework for the combined use of LiDAR-based vehicle trajectory data, microsimulation, and surrogate safety assessment tools to transportation professionals. This integration allows for more accurate and proactive safety evaluations, which are essential for designing safer transportation systems, effective traffic control strategies, and predicting future congestion problems.

A non-stationary bivariate extreme value model to estimate real-time pedestrian crash risk by severity at signalized intersections using artificial intelligence-based video analytics

Vehicle-pedestrian crashes are generally severe due to the vulnerability of pedestrians compared to the occupants of vehicles. However, the estimation of pedestrian crash risk by severity has not been given adequate attention in the field of proactive safety assessments applying traffic conflict techniques. This study proposes a novel analytical framework to estimate real-time pedestrian crash risk by severity at the signal cycle level while incorporating the effect of time-varying exogenous variables. Specifically, the study applies a non-stationary bivariate extreme value model to jointly model the post encroachment time and Delta-V for estimating real-time pedestrian crash risk by severity at individual signal cycles. The proposed framework is tested on 144 h of video data collected from three signalized intersections in Queensland, Australia. The developed bivariate extreme value model has been found to reliably predict severe and non-severe pedestrian crash frequencies compared to the historical crash records of severe and non-severe pedestrian crashes at those signalized intersections. Results suggest that the frequency of pedestrian conflicts per signal cycle and average pedestrian speed in a signal cycle are associated with real-time pedestrian crash risks. In addition, pedestrian conflicts per signal cycle and average vehicle speed per cycle were associated with the interaction severity component of the non-stationary bivariate extreme value model. The proposed proactive estimation of pedestrian crash risk by severity levels can help design time-sensitive countermeasures for vulnerable road users.

A bi-level framework for real-time crash risk forecasting using artificial intelligence-based video analytics.

This study proposes a bi-level framework for real-time crash risk forecasting (RTCF) for signalised intersections, leveraging the temporal dependency among crash risks of contiguous time slices. At the first level of RTCF, a non-stationary generalised extreme value (GEV) model is developed to estimate the rear-end crash risk in real time (i.e., at a signal cycle level). Artificial intelligence techniques, like YOLO and DeepSort were used to extract traffic conflicts and time-varying covariates from traffic movement videos at three signalised intersections in Queensland, Australia. The estimated crash frequency from the non-stationary GEV model is compared against the historical crashes for the study locations (serving as ground truth), and the results indicate a close match between the estimated and observed crashes. Notably, the estimated mean crashes lie within the confidence intervals of observed crashes, further demonstrating the accuracy of the extreme value model. At the second level of RTCF, the estimated signal cycle crash risk is fed to a recurrent neural network to predict the crash risk of the subsequent signal cycles. Results reveal that the model can reasonably estimate crash risk for the next 20-25min. The RTCF framework provides new pathways for proactive safety management at signalised intersections.

Driving fatigue increases after the Spring transition to Daylight Saving Time in young male drivers: A pilot study

The Spring transition to Daylight Saving Time (DST) has been associated with several health and road safety issues. Previous literature has focused primarily on the analysis of historical crash and hospitalization data, without investigating specific crash contributing factors, such as driving fatigue. The present study aims to uncover the effects of DST-related circadian desynchrony and sleep deprivation on driving fatigue, by means of a driving simulator experiment.Eighteen participants (all males, age range 21–30 years, mean = 24.2, SD = 2.9) completed two 50-minute trials (at one week distance, same time and same day of the week) on a monotonous highway environment, the second one taking place in the week after the Spring transition to DST. Driving fatigue was evaluated by analysing several different variables (including driving-based, physiological and subjective indices) and by comparison with a historical cohort of pertinent, matched controls who had also undergone two trials, but in the absence of any time change in between.Results showed a considerable rise in fatigue levels throughout the driving task in both trials, but with significantly poorer performance in the post-DST trial, documented by a worsening in vehicle lateral control and an increase in eyelid closure. However, participants seemed unable to perceive this decrease in their alertness, which most likely prevented them from implementing fatigue-coping strategies. These findings indicate that DST has a detrimental effect on driving fatigue in young male drivers in the week after the Spring transition, and provide valuable insights into the complex relationship between DST and road safety.

Crash frequency prediction based on extreme value theory using roadside lidar-based vehicle trajectory data

Crash prediction models (CPMs) are mostly developed using statistical or data-driven methods that rely on observed crashes. However, the historical crash records can be unreliable due to availability and data quality issues. Near-crashes based CPMs offer a proactive approach to predict crash frequencies prior to the occurrence of crashes. Surrogate safety measures can be used to identify near-crashes from road user trajectories. Roadside LiDAR offers an innovative approach to collect vehicle trajectory data at a microscopic resolution with high accuracy providing detailed information of all road user movements. This study presents a methodology to identify near-crashes from Roadside LiDAR based vehicle trajectory data using the surrogate indicators: TTC (Time to Collision), PET (Post Encroachment Time), ACT (Anticipated Collision Time) and MaxD (Maximum Deceleration). Additionally, time-based, and evasive-action-based surrogate measures are combined as different pairs to obtain crash probabilities using extreme value theory (EVT). The study results show that the bivariate EVT model displays a better fit to conflict extremes, predicting crash frequencies better than the univariate model. Likewise, while the bivariate model with ACT and MaxD pair performed the best in terms of accuracy, the TTC and MaxD pair was able to reflect the relative threat levels at the study intersections. Overall, the methodology lays ground for using roadside lidar based trajectory data for proactive safety analysis of signalized intersections.

Corridor-level network screening and modeling of fatal and serious injury crashes on urban and suburban arterial corridors in Florida

Network screening is a key aspect of the safety management process. The current standard methods in the Highway Safety Manual assume that intersections and roadway segments are independent, even though interactions between sites likely exist. One way to account for these interactions is by studying roadway corridors made of consecutive intersections and segments. This paper developed a roadway corridor definition to identify urban and suburban arterial corridors in Florida with context classification and lane count as defining features. Applying this definition to 10 Florida counties resulted in 1105 corridors with 14,097 fatal or serious injury crashes from 2017 to 2021. Historic crashes and a developed regression model were used to estimate the expected crash frequency for fatal and serious injury crashes with the empirical Bayes method. After performing k-means clustering on expected crash frequency per mile, 25 high-risk corridors were identified. The difference between the expected and predicted crash frequencies (potential for safety improvement) for each high-risk corridor provided a list of priority corridors for safety improvements. Countermeasures focused on speed reduction in suburban corridors would likely lead to the greatest reductions in fatal and serious injury crashes. This corridor approach allows agencies to quickly identify high-risk areas for priority treatment.

A Review of Surrogate Safety Measures Uses in Historical Crash Investigations

Historical road crash data are the main indicator for measuring road safety outcomes. Over the past few decades, significant efforts have been made in obtaining and exploiting Surrogate Safety Measures (SSMs). SSMs have the potential to provide excellent sustainable road safety indicators and proxy measurements which can complement traditional historical crash analyses or even substitute them. By using SSMs, crash data collection demands can be bypassed and areas can be investigated before crashes occur. Due to such advantages, the objective of the present research is to provide a review of the scientific literature regarding studies exploiting SSMs for historical crash record investigations. Specifically, 34 studies were examined, providing insights on the different types of SSMs collected under real road environment conditions, the way they are collected, their connection with specific road crash types, and the type of the developed statistical models are examined and discussed. Particular focus is also placed on the temporal dimension of the collection period of both SSMs and road crashes. Finally, the overall trends deriving from the reviewed studies are summarized and future research directions are provided.

Estimating real-time crash risk at signalized intersections: A Bayesian Generalized Extreme Value approach

The current reactive road safety assessment cannot assess real-time crash risk at signalized intersections, and as such, the real-time risk mitigation strategies could not be developed. This shortcoming is mainly attributed to a lack of a proper methodological framework that could reveal insights into the real-time crash risk using the crash signatures or precursor of crashes, otherwise known as traffic conflicts, near misses, or surrogate safety measures. This study proposes a traffic conflict-based crash estimation technique to estimate real-time crash risk at signalized intersections. In particular, a Bayesian modeling framework is employed to estimate crash risk in real-time from traffic conflicts identified by modified time-to-collision (MTTC). A Block Maxima approach corresponding to generalized extreme value distribution is used to identify traffic extremes at a micro (i.e., traffic signal cycle) level. Then, the traffic flow at the signal cycle level is included as a covariate in the model to explain time-varying crash risk across different cycles. The unobserved heterogeneity associated with the crash risk of different cycles is also addressed within the Bayesian framework. The proposed framework is tested using a total of 96 h of traffic movement video data from a signalized intersection in Queensland, Australia. A comparison between the estimated crashes and the historical crash records demonstrates the suitability of the developed model for crash risk estimations. The crash risk of each signal cycle is identified by generating a different generalized extreme value distribution for each traffic signal cycle. Further statistical analyses reveal that crash risk varies within the different periods of a typical day, with higher crash risks found during the morning and evening peak periods. This study concludes the efficacy of the proposed real-time framework in estimating the rear-end crash risk at the micro-level, allowing proactive safety management and the development of risk mitigation strategies.

Developing Pedestrian Fatality Prediction Models Using Historical Crash Data: Application of Binary Logistic Regression and Boosted Tree Mechanism

Pedestrian fatality rate plays a key role in examining effectiveness of the road safety. The present study attempts to examine the effect of various categories of accused vehicles and the average 85th percentile speed at accident location on the pedestrian crash fatality. The study also attempts to develop pedestrian crash severity models using the binary logistic regression and boosted trees technique. Historical crash data, along with the video recording technique at accident sites, have been utilized for the present study. From regression equations, it is observed that when the heavy vehicle (HV) hits a pedestrian as compared to two-wheeler (2W), the average chance of death increases 2.44 times. According to the Boosted tree model, the contribution of speed is 60 %, whereas the contribution of category of accused vehicle is 40 % for pedestrian fatality prediction. The study should help in planning better strategies like all red time at intersections or pedestrian foot over bridge at critical locations.

Exploring Crowdsourced Hard—Acceleration and Braking Event Data for Evaluating Safety Performance of Low-Volume Rural Highways in Iowa

There are over four million miles of two-lane roadways across the United States, of which a substantial portion is low-volume roads (LVR). Traditionally, most traffic safety efforts and countermeasures focus on high-volume high-crash urban locations. This is because LVRs cover an extensive area, and the rarity of crashes makes it challenging to use crash data to monitor the safety performance of LVRs regularly. In addition, obtaining up-to-date roadway information, such as pavement or shoulder conditions of an extensive LVR network, can be exceptionally difficult. In recent times, crowdsourced hard-acceleration and braking event data have become commercially available, which can provide precise geolocation information and can be readily acquired from different vendors. The present paper examines the potential use of this data to identify opportunities to monitor the safety of LVRs. This research examined approximately 12 million hard-acceleration and hard-braking events over a 3-months period and 26,743 crashes, including 9373 fatal injuries over the past 5-year period. The study found a moderate correlation between hard acceleration/hard-braking events with historical crash events. This study conducted a hot spot analysis using hard-acceleration/hard-braking and crash datasets. Hotspot analysis detected spatial clusters of high-risk crash locations and detected 848 common high-risk sites. Finally, this paper proposes a combined ranking scheme that simultaneously considers historical crash events and hard-acceleration/hard-braking events. The research concludes by suggesting that agencies can potentially use the hard-acceleration and hard-braking event dataset along with the historical crash dataset to effectively supervise the safety performance of the vast network of LVRs more frequently.

Leveraging Social Media Data to Understand Spatial and Severity of Roadway Crashes in Tanzania

Traffic crashes are among the major cause of fatalities in most developing countries. A lack of a centralized historical crash database impedes the efforts to perform large-scale studies to understand the causes of crashes. However, the information for most major crashes are normally shared on social media. Therefore, this study used social media posts to extract information related to head-on and run-off roadway (ROR) crashes that occurred between 2010 and 2020. A total of 661 crashes were collected, which included 409 head-on and 252 ROR crashes. Geographical Information System (GIS) applications and multinomial logit models were developed to understand the spatial distribution of crashes and the resulting severity. It was found that the spatial distribution of crashes differs significantly by type and severity. Head-on collisions were predominant in two eastern regions, while ROR collisions appeared to spread on a large scale across the entire country. The pattern of fatal collision includes one southern region in addition to the two Eastern regions. Conversely, the pattern of non-injury crashes shows predominance along the coastal area. The Multinomial model results showed that speeding was two times more likely to result in fatalities for ROR crashes. Crashes in rural areas were about two times more likely to be fatal, given that they were head-on collisions. Further, bus-involved crashes were less likely to result in injuries. The implications of these study findings to the practitioners are discussed.

Machine Learning Framework for Real-Time Assessment of Traffic Safety Utilizing Connected Vehicle Data

Assessment of roadway safety in real-time is a necessary component for providing proactive safety countermeasures to ensure the continued safety and efficiency of roadways. A framework for utilizing data from connected vehicles and other probe sources is proposed in this study. Connected vehicles present an opportunity to provide live fingerprinting and activity monitoring on roadways. Taking advantage of high-resolution trajectory data streaming directly from connected vehicles, variables are extracted and the relationship with crashes are explored utilizing statistical and machine learning models. Hard acceleration events, in conjunction with segment miles are shown to have strong positive correlations with historical crash outcomes as proven by OLS, Poisson and Gradient Booster regression models. An XGBoost classification model is then trained to predict the real-time instances of crash outcomes at 5 min temporal bins with high levels of accuracy when trained with data including the real-time segment speed, reference speed, segment miles, a segment crash risk factor and other variables related to the difference in speeds between consecutive segments as well as the hour of the day. A weighted ensemble model achieved the best performance with an accuracy of 0.95. The results present evidence that the framework can capitalize on the richness of data available via connected vehicles and is implementable as a component in Advanced Traffic Management Systems for the analysis of safety critical situations in real-time.

Benefit–Cost-Based Method to Determine When Safety Performance Functions Should be Redeveloped for Use in Intersection Network Screening

Network screening is the process of identifying the sites within the road network that are of highest priority for detailed safety audits and interventions. Contemporary road safety network screening methods rely on safety performance functions (SPFs), which are statistical models relating site characteristics to the number of crashes. SPFs are developed on the basis of historical crash and traffic volume data and should be periodically redeveloped to accurately reflect changes in the underlying relationships that result from changes in the network, traffic behavior, operational performance of vehicles, rules of the road, and other contributing factors. However, presently, there is no guidance available to practitioners to enable them to objectively decide when redevelopment of SPFs is warranted. In this study, we present a practical method to provide this guidance for network screening of intersections that reflects both the expected benefits of SPF redevelopment in relation to improved network screening accuracy and the cost of redevelopment. The data required for this method are either already available in municipalities which have locally developed SPFs or can be easily estimated. The estimation models for the method are created based on real data sets and were successfully tested on validation data sets.

Modeling the accuracy of traffic crash prediction models

Crash forecasting enables safety planners to take appropriate actions before casualty or loss occurs. Identifying and analyzing the attributes influencing forecasting accuracy is of great importance in road crash forecasting. This study aims to model the forecasting accuracy of 31 provinces using their macroeconomic variables and road traffic indicators. Iran's road crashes throughout 2011–2018 are calibrated and cross-validated using the Holt-Winters (HW) forecasting method. The sensitivity of crash forecast reliability is studied by a regression model. The results suggested that the root mean square error (RMSE) of crash prediction increased among the provinces with higher and more variant average monthly crashes. On the contrary, the accuracy of crash prediction improved in provinces with higher per capita GDP, and higher traffic exposure. A 1% increase in crash variability, average historical crash count, GDP per capita, and traffic exposure, respectively, resulted in a 0.65%, 0.52%, −0.38%, and −0.13% change in the RMSE of forecasting. The addition of traffic exposure and macroeconomic factors significantly enhanced the model fit and improved the adjusted R-squared by 14% compared to the reduced model that only used the historical average and variability of crash count as the independent variables. The findings of this research suggest planners and policymakers should consider the notable influence of macroeconomic factors and traffic indicators on the crash forecasting accuracy.

Analysis of Surrogate Safety Performance Parameters for an Interurban Corridor

Traditionally, road safety performance evaluation is an analysis of crash data from the past. However, methods of analysis from crash data have some well-known limitations from an analytical point of view. These limitations include small data samples causing statistical issues in analysis, under reporting of crashes and very little information about behavioral and environmental aspects at crash time. The micro simulation combined with traffic conflict technique enables the transportation engineers to investigate the safety performance of a corridor without using the crash data. Surrogate Safety Assessment Model (SSAM), utilizes simulated vehicle trajectories from the micro simulation software to investigate conflict severity and locations. In this study, safety performance evaluation is carried out of an interurban corridor of 24.3 km stretch from Gurugram to Faridabad in state of Haryana (India) using simulation software VISSIM (Verkehr In Städten – SIMulationsmodell) and SSAM. Simulated vehicular trajectories were generated and analyzed using SSAM to identify potential conflicts. The surrogate safety measures Time to Collision (TTC), Post Encroachment Time (PET) and Max ΔV are obtained by an analysis from SSAM model for all the three homogeneous sections such as midblock, curve section and intersections separately. The approach presented in the paper helps in the identification of inter-urban corridor locations prone to road crashes and hence serves as a proactive alternative as opposed to historical crashes based analyses.

Assessment of Crash Occurrence Using Historical Crash Data and a Random Effect Negative Binomial Model: A Case Study for a Rural State

This work identifies factors that influence crash occurrence within a traffic analysis zone (TAZ) by accounting for location-specific effects and serial correlation in longitudinal crash data. This is accomplished by applying a random effect negative binomial (RENB) model. Unlike commonly used count models such as Poisson and negative binomial (NB), RENB accounts for heterogeneity and serial correlation in crash occurrence. An RENB was applied to 15 years of crash data in Arkansas with 1,817 TAZs. Four models were developed for total crashes and by severity (property damage only (PDO), injury, and fatal). RENB-estimated impacts were measured using the incidence rate ratio (IRR). The significant causal factors found to increase in observed crashes include: (i) average precipitation (a one-unit increase in average precipitation results in a 134% increase in total monthly crashes for a TAZ); (ii) average wind speed (16%); (iii) urban designation (7%); (iv) traffic volume (2%); and (v) total roadway mileage (1% for each functional class). Snow depth and days of sunshine were found to decrease the number of accidents by 15% and 2%, respectively. Employment and total population had no impact on crash occurrence. Goodness-of-fit comparisons show that RENB provides the best fit among Poisson and NB formulations. All four model diagnostics confirm the presence of over-dispersion and serial correlation indicating the necessity of RENB model estimation. The main contribution of this work is the identification of crash causal factors at the TAZ level for longitudinal data, which supports data-driven performance measurement requirements of recent federal legislation.

Massive GNSS data for road safety analysis: Comparing crash models for several Canadian cities and data sources

Mobile sensors are a useful data source with applications in several transportation fields. Though cost of collection, transmission, and storage has limited studies on driving data and safety, this can be overcome through usage-based insurance (UBI). In UBI programs, drivers are monitored, and their premiums are adjusted based on driver-level surrogate safety measures (SSMs) related to exposure and driving style. Contextual link-level SSMs (volume, speed, or density) could further improve discount calibration. This study quantifies relationships between contextual SSMs and crashes and includes the validation of previous results (correlations between SSMs and crashes and statistical models estimated using smartphone-collected data from Quebec City) and the comparison of three Canadian cities (using UBI data from Quebec City, Montreal, and Ottawa). Extracted SSMs were compared to large volumes of historical crash frequency data using Spearman’s Rank Correlation Coefficient and then implemented into spatial Bayesian crash models. Results from the UBI data generally matched those from the previous study, with observed correlations mirroring previous results in direction (braking, congestion, and speed variation are positively associated with crash frequency while mean speed is negatively associated) while correlation strength was slightly higher. Furthermore, these results were consistent between cities. For the crash modelling, repeatability of previous results in Quebec City was moderately good for the UBI data. Importantly for large-scale implementation, models estimated using UBI data were largely consistent between cities. This work provides an important contribution to the existing literature, clearly demonstrating how contextual safety measures could be applied to benefit UBI practices.

Evaluation of hydroplaning potential using Mobile Lidar measurements for network-level pavement management applications

The majority of weather-related crashes in the U.S. occur on wet pavement, potentially due to hydroplaning. However, hydroplaning potential is often not considered in network-level pavement condition assessments. This paper provides a method for evaluating hydroplaning potential based on the actual road surface and geometric properties measured using mobile Light Detection and Ranging (lidar). The lidar measurements were placed in a grid format, and water film thickness was determined based on water depth and pavement macrotexture. A Monte Carlo simulation produced a travelling speed at which hydroplaning could occur. The developed method was applied to an in-service roadway section with historical wet weather crashes. It successfully identified the critical drainage basin and determined that the travelling speed at which hydroplaning could occur is lower than the posted speed limit. This method can be integrated into network-level pavement management systems to identify roadway sections with high risk of wet weather crashes.

Diverging Diamond Interchange Performance Measures Using Connected Vehicle Data

Since the first Diverging Diamond Interchange (DDI) implementation in 2009, most of the performance studies developed for this type of interchange have been based on simulations and historical crash data, with a small number of studies using Automated Traffic Signal Performance Measures (ATSPM). Simulation models require considerable effort to collect volumes and to model actual controller operations. Safety studies based on historical crashes usually require from 3 to 5 years of data collection. ATSPMs rely on sensing equipment. This study describes the use of connected vehicle trajectory data to analyze the performance of a DDI located in the metropolitan area of Fort Wayne, IN. An extension of the Purdue Probe Diagram (PPD) is proposed to assess the levels of delay, progression, and saturation. Further, an additional PPD variation is presented that provides a convenient visualization to qualitatively understand progression patterns and to evaluate queue length for spillback in the critical interior crossover. Over 7000 trajectories and 130,000 GPS points were analyzed between the 7th and the 11th of June 2021 from 5:00 AM to 10:00 PM to estimate the DDI’s arrivals on green, level of service, split failures, and downstream blockage. Although this technique was demonstrated for weekdays, the ubiquity of connected vehicle data makes it very easy to adapt these techniques to analysis during special events, winter storms, and weekends. Furthermore, the methodologies presented in this paper can be applied by any agency wanting to assess the performance of any DDI in their jurisdiction.