International Conference on Transportation and Development 2020: Transportation Safety

This study conducts geometric and operational variable investigation to analyze motorcycle-related crashes on Interstate 65 (I-65) in Kentucky using three large metropolitan cities as case studies. These are the Cities of Louisville, Bowling Green, and Elizabethtown. Five-year motorcycle crashes (from 2014 to 2018), exported from the Kentucky State Police Collision Database, were used in the analysis. Rarely-explored roadway geometric and operational variables (collected using Google Maps) were used in the motorcycle crash investigation, such as: left shoulder width, lane width, number of lanes, median type, median width, and crash distance to the nearest on/off ramp. The analysis showed that the majority of motorcycle-related crashes (71%) occurred in Louisville. Most crashes occurred on I-65 roadway segments with fewer numbers of lanes (i.e., one to three lanes), narrower lane widths (10 to 12 feet), and wider medians (>14 ft) divided by concrete barriers. Overall, lower motorcycle injury percentage (63%) occurred at distances less than or equal to 1,145 ft (about 0.22 mi) from the on/off ramp, as opposed to distances greater than 1,145 ft (76%). A high motorcycle injury percentage (95%) occurred on wider lanes (>12 ft), as opposed to narrower lanes (10–12 ft), mostly due to speeding. In multiple-vehicle crashes between motorcycle(s) and other vehicle(s), motorcyclists suffered the most severe crash injuries, as opposed to the other involved vehicle drivers. The factors that were determined to significantly affect motorcyclists’ crash injury severity based on the chi-square test of independence were: helmet use, season of the year, weather conditions, lighting conditions, lane width, number of lanes, median width, crash distance to the nearest on/off ramp, and motorcyclist pre-collision action. The Z-test of proportions showed significant motorcycle crash proportion difference between Louisville and Elizabethtown concerning divided concrete barrier median types (65.3% in Louisville versus 22.2% in Elizabethtown). Some safety recommendations included increased helmet use for all motorcyclists, increased educational programs on motorcycle safety for both motorcyclists and vehicle drivers, and the possibility of dedicating a separate travel lane for motorcyclists on those high-crash risk segments on I-65.

Determining the conditional probability of a crash with a narrow-fixed object (NFO) given an encroachment has occurred, P(C|ENCR) has been computationally burdensome leading to the development and reliance on software by roadside safety engineers. The objective of this study was to develop a simple model to represent P(C|ENCR) when the size of and the offset to an NFO are known. This paper presents a model developed from the field collected data available within the third version of the roadside safety analysis program (RSAPv3) to represent P(C|ENCR) for individual NFOs. This model is an equation which can be applied to NFOs across a variety of diameters and offsets from the travel way. The use of this model will allow for quick computations to determine the probability of a crash with an NFO across a variety of offsets and diameters without the necessity to run software. Models are presented for both heavy vehicles and passenger vehicles. The methodology, datasets, and analysis used are documented. This research led to an improved understanding of the influence of offset and diameter on the probability of a crash. Models such as these reduce the dependence on computationally intense software programs and allow the analysis of safety to be more transparent.

One important aspect of the selection of the longitudinal barrier for use in a particular situation is estimating the probability that vehicles will cross the barrier line by penetrating, rolling over, or vaulting over the barrier. This probability is a function of the barrier type and the expected vehicle types that will interact with the barrier. This paper presents a method based on a semi-empirical approach using observed crash data, reconstructed field trajectories, a simplified mechanics model, and RSAPv3 simulations to estimate the probability of passing through a barrier based on the barrier types and vehicle type. This information is necessary when selecting the appropriate MASH test level barrier for a particular situation based on the traffic conditions expected at the site.

Roadside safety features generally demonstrate successful performance through crash testing, which provides an assessment of occupant risk, post-impact trajectory, and structural adequacy. It has been long recognized that the performance of safety features cannot be assessed by crash testing alone and should be followed by in-service performance evaluations (ISPEs). The primary objective of any ISPE is to evaluate occupant risk under real-world field conditions. A secondary objective may be to determine which factors influence performance (e.g., maintenance, installation, hardware design, etc.). This paper assembled available data from both the literature and the states of Ohio, Pennsylvania, and Tennessee to demonstrate the various data sources readily available to answer the primary ISPE question of occupant risk. While much more can be done with the data available from each of these states, this paper is limited to a review of occupant risk. This paper has shown that there is a very large amount of data very easily accessible for an ISPE. While some states maintain an inventory of safety features, some do not, and additional information was needed to identify the feature and proceed with the evaluation.

Despite the recent efforts to investigate crash severity, worker presence and its impact on injury severity in work zone crashes is still unexplored. A better understanding of work zone crash characteristics can help to enhance roadway safety for not only road users, but construction crew. Employing a mixed logit (MXL) modeling framework, the present study aims to identify and investigate contributing factors associated with work zone crashes involved workers. Random forest (RF), a data mining approach, is also applied to evaluate variables’ importance for comparison between worker-involved and non-worker involved crashes. The estimation results demonstrated that work on the shoulder or median, advance warning area, daytime non-peak, and multi-occupant variables have heterogeneous effects on injury severity. The analysis of marginal effects indicated variables with substantial influences on injury severity outcomes. The results of this study provide valuable insights for work zone crashes analysis.

Rural intersection safety continues to be a crucial issue throughout the United States. More than 20% of all traffic fatalities in the United States occur at intersections, and over 80% of intersection-related fatalities in rural areas occur at unsignalized intersections. This paper investigates the significant factors affecting crash severity at stop-controlled intersections on rural two-lane highways in Louisiana. Roadway characteristics, traffic characteristics, and crash characteristics were collected from 5,136 stop-controlled intersections including both three-leg (3T) and four-leg (4T) of all 64 Parishes (counties) statewide, for the period of 2013 to 2017. Zero-inflated Poisson (ZIP) models were developed to model the number of intersection crashes for different severity levels. In addition to traffic volume, horizontal curve radii, speed limits of major roads, and intersection skewness also have significant impact on crash occurrences.

Civil aviation safety is one of the crucial problems of the aviation industry. Air traffic management as a core part of civil aviation is facing massive pressure of safety operation. Accurately evaluating unsafe events rate is essential to improving air traffic safety. This paper proposed a grey-buffer operator-Markov chain method to forecast the civil aviation unsafe events rate. It combined the grey system and Markov chain. When predicting the civil aviation unsafe events rate, the actual data samples are few, discrete, and random fluctuation, which often causes more significant relative errors and worse accuracy using the traditional grey model. However, the average weakening buffer operator is an excellent method to process the raw data. According to the distribution of relative value, we can build the prediction model, combining with the characteristics of Markov forecast volatile data sequence. Finally, this method is applied to predict the unsafe events rate of civil aviation. The experimental result shows that this method not only reduces the relative error of predicted value but also improves the prediction precision of the model.

More than 80% of the traffic information acquired by drivers comes from the visual channel. Therefore, visual information provided by traffic environment has great effects on driving safety. Data of drivers’ driving trajectories, driving velocities, and eye movement characteristics in experimental scenes with different traffic environmental visual information and levels of self-explaining characteristic were collected via a driving simulator and eye tracker experiment. This paper analyzed the effects of visual information on drivers’ behaviors, evaluated the safety of self-explaining roads, and designed the methods to improve the capacity of fault tolerance and safety of the self-explaining intersection.

Intersections are sites where traffic crashes are highly concentrated. This is because conflicting traffic maneuvers are highest at the intersections and, therefore, they are more prone to crash occurrence. An accurate understanding of the relationships between various geometric features, traffic flow and crashes in the influence area of intersections, could help reduce these crashes by suggesting appropriate countermeasures. Safety performance functions are the tools to understand these relationships. In this study, we developed safety performance functions specific to Antwerp, Belgium, for urban intersections, using multiple functional forms of a Poisson-gamma framework considering total crash counts. A dataset consisting of six years (2010–2015) of intersection crashes, traffic flows (major and minor), and geometric features was created. This paper documents the development process of safety performance functions for signalized and un-signalized intersections, connecting roadway geometry and traffic characteristics to intersection related crashes and, thus, estimating the impact of predictor variables on the crash occurrence, so as to facilitate the safe design. The findings indicate several road geometric and traffic variables that have a significant impact on crashes at urban intersections. The results revealed that to improve intersection safety, enough attention must be paid to the number of through lanes of a minor approach, intersection skewness, size, and presence of crosswalks on the approach roads. Furthermore, it was found that an early inquiry into the model’s functional form could result in more accurate estimates of crash frequency.

This paper documents the calibration of the highway safety manual (HSM) safety performance functions (SPFs), quality assessment of calibrated HSM SPFs, and development of calibration functions for crossroad ramp terminals using Kansas crash data from 2014 to 2016. As SPFs in the HSM were developed using data from several states, these crash prediction models must be calibrated to improve the crash prediction accuracy of a locality. For this calibration, 120 stop-controlled ramp terminals and 74 signal-controlled ramp terminals were used. Considering both stop-controlled and signal-controlled crossroad ramp terminals, the HSM methodology underpredicted fatal and injury crashes and overpredicted property damage only crashes in Kansas. Estimated calibration factors for crossroad ramp terminals were satisfactory; however, calibration functions produced better fitted models that improved the fit to local data. The estimated calibration factors and calibration functions greatly benefit in making accurate decisions related to ramp terminal safety in Kansas.

While talking about road maintenance and safety in mobile work-zones, it is unavoidable to not talk about truck mounted attenuators (TMAs). TMAs are eminent in work zones as they are mainly intended to reduce damage in case of vehicular crashes and at the same time protect construction workers at the work-zones. In this paper, here followed a differential analysis approach to obtain a macro view of TMAs-involved crashes and considered possible countermeasures. This research used TMAs-related crash data recorded in the U.S. state of Missouri from 2011 to 2016 and referred to both crash and claim reports for current analysis. A new approach for studying TMAs-involved crashes has been proposed with several variables being evaluated based on existing work zone crash prediction models. The proposed method could be useful at finding the staple source of traffic accidents and potential safety hazards, provide the main references for TMAs work safety protection.

On July 1, 2013, the Ohio’s legislature raised speed limits on rural freeways from 65 to 70 mi/h for all vehicles. This study investigated the impact of raising speed limits on crash severity specifically with interest on heavy vehicles on rural freeways in Ohio. Statewide crash data from January 1, 2010, to December 31, 2018, were used in this study. Because of the model requirement for stationarity on a response series, this study utilized the autoregressive integrated moving average (ARIMA) time series intervention analysis method using monthly crash data. Results of the study show that raising speed limits on selected segments of Ohio rural freeways has an impact on the frequency and severity of crashes associated with large trucks and buses. Moreover, the study found that bad winter weather conditions had a significant impact on the frequency and severity of crashes associated with large trucks and buses as they happen to be higher during winter seasons as compared to other times of the year.

Various daily driving habits and styles may be the reason why drivers show individual differences when taking over. The purpose of this paper is to investigate the impact of driving style on take-over performance in Level 3 automation. The simulator experiments in this study consisted of two parts: a manual car-following experiment and an automated driving take-over experiment. In the manual car-following experiment, researchers used the data of maximum braking deceleration, and maximum acceleration to classify 20 participants into two driving types (12 normal drivers, 8 aggressive drivers). Then the automated driving take-over experiment was conducted to investigate the influence of driving styles (normal, aggressive) on take-over performance in different take-over time budgets (7 and 5 s) and in combination with a visual secondary task. The researchers found that normal drivers took a significant severer brake than aggressive drivers only in the condition of 5 s take-over time budget + SuRT (i.e., the most urgent of all conditions). In addition, the driving style had no significant influences on take-over time, minimum TTC, and maximum lateral acceleration. Results indicate that normal drivers’ good habits do not make their take-over performance better, while the experience of aggressive drivers in manual emergency driving makes their emergency evasive maneuvers more stable in automated take-over.

Vehicle conflicts in freeway weaving sections occur when vehicles are required to make lane changes, while operating in a confined space, without the aid of traffic control devices. The safety issue associated with lane changes is that drivers must remove their attention from the vehicle they are following and focus on vehicles next to and behind them while continuing to drive at a high speed. This change in focus has the potential to lead to crashes at a much higher rate than would be expected during normal freeway driving conditions. The objective of this paper is to analyze driver, environmental, roadway, geometrics, and operational characteristics to determine why crashes occurred and who is more likely to have crashes in weaving sections. The results revealed that 58.5% of weaving crashes were rear-end crashes on freeway weaving sections.

Posted speed limit (PSL) on most of the Interstate highway system in Wisconsin was increased from 65 to 70 mph during the week of June 16, 2015. This paper aimed at evaluating the impact of the change in speed on Wisconsin’s Interstate system after the maximum PSL increased. The primary data source was the national performance management research data set (NPMRDS). One-week time periods every month from March 2014 to May 2015 were collected for the before period. One-week time periods in every month from July 2015 to June 2016 were collected for the before period.

Pedestrian safety at intersections, where motor vehicles cross paths with people walking, is a serious matter of concern for traffic and road safety engineers, and professionals. On a nationwide scale, according to the National Highway Traffic Safety Administration, there were 5,977 pedestrian fatalities and more than 85,000 pedestrian injuries as a result of traffic crashes in 2017. This paper provides the results of the observational study to evaluate the effectiveness of the Street Smart NJ pedestrian safety campaign. The effectiveness of the campaign was explored by comparing the rates of non-compliant pedestrian and driver behaviors before and after the campaign. The studied non-compliant behaviors include unsafe crossing and crossing against a signal, failing to stop for pedestrians when turning, failing to stop before turning at a red light or stop sign, and running the red light signal or stop sign. To do so, video data were collected in eight communities across New Jersey in 2018 and 2019. Overall, the results of the study show significant improvements in pedestrian and driver behaviors following the safety campaign.

Wildlife-vehicle collision (WVC) is a major problem associated with regions with high-density wildlife. Urban designers have in the past introduce overpasses, underpasses fence, reflectors, and sensors to aid safe wildlife road crossing, but these have not been able to reduce the wildlife-vehicle collision. This research focused on the automated warning system to vehicle users to minimise wildlife-vehicle collision which could integrate computer vision in the detection of features on the road together with the location-time information feed. The proposed system was trained using AlexNet, GoogelNet, ResNet-50, and VGG-16 algorithm on a deep convolutional neural network (CNN) using 20,964 images of 25 variables consisting of 21 animals and four different vehicle body type. The dataset was divided into training and validation set. The results show that CNN algorithms could identify objects on real-life traffic data with noise background at a reliable accuracy. The GoogelNet, ResNet-50, and VGG-16 model outputs were found to have a better prediction accuracy than the AlexNet model in detecting the object features on the traffic images.

In the U.S., total 94,000 crashes (Fatal 710, Injury 36,290, and PDO 57,000) occurred in work zones in 2017. The economic cost to these crashes is estimated at nearly $17.5 billion. Data revealed that there is a greater risk of injury crashes on interstate highways compared to other functional classification of roadways. In order to improve work zone safety on interstate highways, this study proposed an optimization methodology that basically identifies a best subset of safety countermeasures in each work zone of a network. The subset of safety countermeasures was selected based on several factors including crash modification factors, safety benefits, traffic counts, implementation costs, benefit-to-cost ratio (BCR), and most importantly limited budget. Linear integer optimization technique was utilized to identify the best subset of safety countermeasures. The methodology was implemented successfully on a dataset to verify the applicability of the model.

Distracted driving is known to be one of the core contributors to crashes in the US, accounting for about 40% of all crashes. Drivers’ situational awareness, decision-making, and driving performance are impaired due to temporarily diverting their attention from the primary task of driving to other tasks not related to driving. Detecting driver distraction would help in adapting the most effective countermeasures. To find the best strategies to overcome this problem, we developed a Bayesian Network (BN) distracted driving prediction model using a driving simulator. In this study, we use a Bayesian Network classifier as a powerful machine learning algorithm on our trained data (80%) and tested (20%) with the data collected from a driving simulator, in which the 92 participants drove six scenarios of hand-held calling, hands-free calling, texting, voice command, clothing, and eating/drinking on four different road classes (rural collector, freeway, urban arterial, and local road in a school zone). Various driving performances such as speed, acceleration, throttle, lane changing, brake, collision, and offset from the lane center were investigated. Here we investigated different optimization models to build the best BN in which a Genetic Search Algorithm obtained the best performance. As a result, we achieved a 67.8% prediction accuracy using our model to predict driver distraction. We also achieved a 62.6% true positive rate, which demonstrates the ability of our model to correctly predict distractions.

The railroad industry plays a crucial role in stimulating the economic growth of a nation. Therefore, monitoring the conditions of railroad is vitally essential for ensuring railroad safety. Railroad maintenance systems have integrated advanced technologies to help inspectors detect and localize anomalies that can cause casualties and financial losses. However, railroads cannot afford to use these complex technologies to monitor and inspect all railroad tracks or important routes more often. Intelligent transportation systems (ITS) embed information, communication, and sensor technologies into vehicles and transportation infrastructure to supply real-time information that agencies can use to make better decisions. This research used an ITS approach to develop a scalable and affordable track condition monitoring system. This paper describes the architecture of the system that will autonomously identify track anomalies to focus follow-up inspections. Data measured from a regional railroad showed that the proposed system would be effective and affordable for practical applications.

In many developing cities with impropriate urban planning strategy and poor drainage system always suffer waterlogging problem during heavy rains. Waterlogging, as a well-known urgent problem for urban development, will increase risk of traffic accidents, cause malfunction of transportation system, and seriously threaten people lives. However, those harm can be effectively reduced by giving an early warning to citizens after detecting the water accumulation. Some existing methods based on detective sensor are proved to be expensive and low precision. Other methods try to apply macro simulation model to predict result of waterlogging, but complicated mathematical modeling, as well as getting access to geographic information and related data sources, is hard to implement. Therefore, this paper proposes an effective data-based approach for waterlogging recognition built on fundamental diagram (FD) of traffic flow. The methodology implements a real-time intelligent recognition on waterlogging points by observing shift on FD and calculating the probability of waterlogging. Compared to existing methods, it has the better applicability on entire urban road network, the easier realizability on application implementation, and more independent on data requirement. A study case of Shenzhen (a city in southern China) waterlogging indicates the methodology can effectively provide a real-time recognition on road segments prone to water accumulation during rainstorm, and the accuracy is proved to be better than any other similar approaches.

For the risk characteristics of the two-lane highway driver, a visual alignment information calculation model is established, based on driving dynamic vision theory, and comprehensively considering the alignment information of horizontal, vertical, cross-section of highways and drivers’ visible area. The relationship between visual alignment information and accident rate was analyzed by the alignment data and accident data of the two-lane highway in Xiangxi, Hunan, China. The result proved that the visual alignment information was poorly correlated with the accident rate and the change rate of visual alignment information had a good positive correlation with the accident rate. The greater the change rate of visual alignment information was, the higher the accident rate was. The accident rate value corresponding to the cumulative frequency of 25%, 50%, 75%, and 90% obtained by the cumulative frequency curve method was combined with the relationship model between the accident rate and the change rate of the visual alignment information to construct a quantitative standard for alignment driving risk of two-lane highway based on the change rate of visual alignment information. Findings of the research could provide theoretical basis for exploring the research on subjective and objective safety of highway and safety assessment.

As a supporting facility, expressway service area is of great significance to the rapid development of expressway. In order to evaluate the effectiveness of traffic signs and markings of expressway service area from human factor, this paper adopts driving simulator technology. The traffic signs and markings of Shangdian expressway service area of Beijing Xingyan expressway was the research object and the immersive 3D driving simulation scenario was built. Then, the driving simulation experiment was designed based on the way-finding theory. 31 subjects were recruited to participate in the experiment. Finally, based on driving behavior data, the effectiveness of traffic signs and markings in each zone was evaluated. The result shows that longitudinal and lateral deceleration marking, sign and uncolored pavement word, colored pavement word should be set in the entrance zone, diverging area for large vehicle and light-duty vehicle and gas station respectively.

In the grassland highway environment, monotonous landscape is likely to cause drivers driving fatigue, which leads to the occurrence of traffic accidents. For this reason, the application of driving fatigue prevention safety facilities is particularly important. In this paper a kind of colored road pavement fatigue prevention safety facility was mentioned. Studies have shown that driving fatigue is easy to cause changes in drivers’ driving behavior and alertness. In this paper, a driving simulation experiment was conducted, based on the analysis of driver’s driving behavior, KSS value and reaction time, significance analysis were used to analyze the effectiveness of the fatigue prevention safety facility. The result showed that the colored road pavement fatigue prevention safety facility could make a difference on drivers’ driving behavior and reduce the increase of drivers’ feelings of fatigue and reaction time. The experimental results could provide reference for the application of fatigue prevention safety facilities.

This study analyzes bicycle-vehicle crashes in Alabama to identify bicycle-vehicle crash causes/patterns. The analysis is based on 1,311 crashes in the state, representing five-year crashes (2011–2015). Detailed review of 1,311 police reports was performed and geometric/roadway variables were collected from Google Maps that were rarely explored in previous bicycle safety studies, such as: type of median on major road and crash location distance to upstream signalized intersection. This study adds to the literature by examining the effect of geometric/roadway features (e.g., presence of bicycle lanes, number of through lanes, and median type) on prioritizing bicycle routes while accounting for “bicycle mileage” exposure. Nighttime riding (19:01–7:00), and crashes involving males and middle-aged individuals (26–64) resulted in more severe bicycle-vehicle crashes. Compared to intersections, more severe crashes occurred at midblock sections at speeds higher than 35 mph. The chi-square test revealed significant correlation between crash severity and each of bicyclist safety equipment (i.e., helmets), intersection layout, and crash distance to upstream signalized intersection. The study also proposes a procedure for prioritizing hazardous bicycle routes while collecting bicycle exposure (bicycle mileage) using the City of Huntsville as a case study. Twenty-five bicycle routes (totaling around 170 mi) in the city were ranked by crash rate (i.e., crashes per mile) to identify hazardous routes. Bicycle routes on roadways with undivided medians had lower crash rates than those with divided medians. Bicycle routes having bicycle lanes on both sides of the roadway showed 50% lower crash rate than those on only one roadway side. Useful recommendations include: outreach programs emphasizing the importance of wearing reflective clothing at night and providing bicycle lanes on both roadway sides on designated bicycle routes.

A large percentage of drivers like listening to music, radio, or talking to passengers while driving, especially phoning in recent days, even if they use bluetooth earphones or car phone instead of holding the phones by hand. However, such extra cognitive workload will have an impact on drivers’ driving performance. In this study, a driving simulator based experiment was designed to find out drivers’ performance under smooth car-following drive and critical situations. In this experiment, 24 participants (male = 13, female = 11) were required to drive the same city road route twice, with and without extra cognitive workload. Driving performance is measured in driving safety and stability with several indicators. While smooth following the lead vehicle, drivers’ operations are probably not change, however, their risk perception have been increased. Once critical situations happen, extra workload put drivers into a more dangerous and higher risk to collision circumstances. Most of the stability metrics show significant difference between drives with and without extra workload. It is hypothesized that extra workload makes drivers distracted so that they need to adjust velocity and position more frequently.

With the development of intelligent connected network and cooperative vehicle- infrastructure systems, the research of dynamic safety distance is still playing an essential role in CVIS. A short distance between moving cars is dangerous for its incapacity of preventing various types of unexpected situations and potential hazards, meanwhile, a distance which is too long is inefficient in utilizing the traffic flow and road resources. Based on the physical model of automobile braking process and AASHTO (American Expressway and Transportation Association) model, this paper establishes the calculation model of critical safety distance in automobile driving process by analyzing the parameters of automobile braking acceleration, the time when the braking force reaches its maximum and driver’s reaction time. The simulation experiments prove that this model can effectively improve the traffic efficiency of highways and avoid traffic accidents, provided a theoretical reference for the automatic driving system, and the assisted driving system.

Road hazardous segments (HS) identification is valid and useful to improve traffic safety and save life. Traditionally, researchers always use long-term historical crash data with the simulation model to find the HS, which is always a way of being wise after the tragedy. However, this paper promotes a scheme of short-term real-time HS identification scheme based on the traffic inconstancy features. Using accumulative historical floating vehicle data to show and capture the road-segment level of spatial-temporal inconstancy features, then combined with the real-time loop detector data along the same road segments, his paper uses an innovative hybrid model called the standard deviation accident model (SDAM) to identify the three risky levels of HS, suspected, hazardous, and dangerous. A day was broken into 83-hour short-term periods, and the model is fitted by two weeks crash data and validated by 3 days, and achieve the accuracy over 73% for identifying the short-term hazard on each road segment (average length 0.826 km) of total 177 road segments. The success of the SDAM model is mainly due to the well-use of the inconstancy information captured from the long-term historical trajectory data. It was found that the inconstancy of speed and deceleration are close related to the HS identification (with a p-value less than 0.001) significantly. The features of HS road segments are summarized, and the safety improvement suggestions are provided at the end of the paper.

Traffic incident is a big challenge for traffic safety management. The duration of traffic incidents is critical to measure the severity of the incident and essential for improving traffic management abilities. We collected the traffic incident data of major urban roads in Tianjin, China, for consecutive 2 years. Our study conducts a comprehensive analysis of the main factors affecting the duration of the traffic incidents, such as the occurrence time, the road type, the vehicle type, and the number of injured. By using the classical statistical regression model and the several popular machine learning techniques, we try to predict the duration of the traffic incidents and compare the results generated by these models. The results show that the prediction errors of the model are distributed between 30% and 40%. The model established by the machine learning techniques has higher precision than simple linear regression models. Notably, support vector regression performs best among these models. Moreover, the tree-based can be applied for a fast estimation of the traffic incident duration, which deserves application in real-life scenarios.

In the U.S., the risk of vehicle crashes is higher among teens than among any other age group. Most previous studies investigated the effects of demographic differences and nonspatial factors associated with crashes such as gender, age, driving under the influence of drugs or alcohol, the presence of passengers, and distractions. This research was conducted to develop a methodology to identify statistically significant spatial patterns for crashes involving teen drivers aged 15–19 using seven years (2010–2016) of crash data, which included 286 fatal crashes and 72,370 non-fatal crashes using ArcGIS. The results showed that all teen-related crashes were clustered and autocorrelated and high values of crashes were also clustered. Most of the statistically significant hotspots and outliers for crashes were centered in the most populated counties. The results provide a useful indicator for related parties to identify where they can target their resources in order to improve teen driver safety.

Crash reporting is a crucial component of roadway transportation safety analysis. In 2017, new crash report forms were introduced in Wisconsin. If or when law enforcement does not file a crash report, drivers are required to complete a self-reported crash form. The analysis in this paper consisted of quantifying the effect of intervention and accuracy of information. Significant changes in police crash reports were observed in most crash severities after intervention. Driver self-reported crashes showed slow adoption and limited information to accurately map crash locations. Driver self-reported crashes were inconsistent compared to conventional KABCO crash distributions. There was a significant difference in police reported and driver self-reported crashes by jurisdiction. Police reported crashes were predominant in populated urban areas and driver self-reported crashes were prevalent in rural areas. The results of this study provide valuable information for improvement in crash reporting.

This paper aims to contribute to the literature by conducting a causation study for large-truck involved crashes. Detailed records of crashes involving large trucks occurred in the state of Florida between 2007 and 2016 were investigated. A hierarchical framework was developed to identify primary causes and contributing factors to large truck crashes. The framework identified six general categories of potential causes to a crash, including driving error, non-driving error, driver distraction or vision obstruction, vehicle defect, roadway conditions, and weather conditions. The application of the framework was illustrated in two case studies including a highway segment and an intersection with a high concentration of large truck crashes. The findings can be used to better understand the underlying factors contributing to large truck crashes and develop potential countermeasures and policies that target specific roadway segments.

As the key nodes of urban road network, intersections are the most complex and hazardous areas. Self-explaining intersections may improve traffic safety by assisting drivers’ performances and reducing their failures. A simulator-based experiment with 10 designed scenes in different intersections was conducted to collect drivers’ behavioral data of 20 participants. Questionnaire surveys were developed to assess the drivers’ driving load. Driving performances under different intersections were analyzed including the speed choice, drivers’ visual characteristics, and the homogeneity of driving trajectories. The results show characteristics of intersections based on self-explaining road theory. The countermeasures to improve fault tolerance and safety of intersections were proposed.

Car driving is a goal-directed behavior with the driver playing a key role. When the drivers are encouraged or hindered from achieving their goals, they may behave in different emotional states on different scenarios which are well known to impact perception, attention, and maneuvers. A simulator based experiment was designed within which the driver emotions were induced by stimulation of different traffic scenarios. The facial expressions of 21 participants were recorded, and three kinds of emotional states including neutral, impatience/anger, surprise were determined according to facial action coding system (FACS). The effects of different scenarios and emotional states on driving performance with the indicators of driving speed, acceleration, and standard deviation of lateral position were statistically analyzed. The results may contribute to the communication of emotions and planned behavior, as well as the identifications of strategies to reduce aggressive driving.