International Conference on Transportation and Development 2023: Transportation Safety and Emerging Technologies

Automated speed cameras (ASE) and red-light cameras are highly effective tools for reducing overall car crashes. ASE and red-light cameras help reduce most crashes, such as fatal crashes, injury crashes, and property damage crashes worldwide. Millions of people die yearly because of car crashes, and many more are severely injured. Due to the high number of fatal and injury crashes, to reduce deaths, many countries across different continents in the world started enforcing ASE and red-light cameras to mitigate the number of crashes. ASE and red-light cameras are enormously effective measures to reduce the totality of crashes. Unfortunately, despite the proven efficacy, many states in the United States and worldwide prohibit using ASE. Some prohibit red-light and speed cameras, such as New Hampshire, South Carolina, Maine, West Virginia, and Texas. Certain states only prohibit red-light cameras, including Montana and South Dakota, and some states do not allow speed cameras, counting Wisconsin and New Jersey. This paper will examine the reasons behind prohibiting automated speed enforcement technology. Additionally, the study will collect and analyze data from the United States, Canada, and Saudi Arabia regarding the amount each country levies for penalties and fines for ASE violations. Some of the public believe that automated speed enforcement techniques are a method for the government to generate revenue. Others argue that the profits are used to educate the public about traffic safety rules and use it to improve the public infrastructure. This study aims to establish whether the funds from automated speed enforcement fines are used to educate the public and improve infrastructure.

During automated driving in signalized intersections, decisions must be made by autonomous vehicles based on the system dynamics and the measurements provided by the sensors. The system dynamics is often affected by some uncertainty due to unmodeled parts of the system, the discrepancies in modeling the vehicle dynamics, and the adaptive timings of traffic lights according to traffic conditions. The measurements perceived by sensors about the traffic lights and the vehicle position are often noisy due to factors including various illumination conditions, incomplete shapes due to occlusion, very few pixels for detecting distant traffic lights, and motion blurring due to high-speed driving. These sources of uncertainty and noises can directly impact the vehicle’s perception about the states of the traffic light and the vehicle, which can cause running the red light or rear-end crashes from the sudden brakes. In this paper, we propose a hidden Markov model (HMM) representation of the dynamical system that takes into account all sources of uncertainty and noisy measurements. The proposed framework estimates the state of vehicle and traffic light via sequential noisy observations through a particle filtering scheme. The reliability of the proposed framework is demonstrated in the numerical experiments.

This paper provides an estimation of crash cost reductions with lane departure systems and an observation of striping detect rates including experimental orange striping. The results indicate high detect rates with all vehicles tested with delayed detection when striping is worn to less than 20% visible. Both economic and comprehensive crash costs were estimated using current and projected technology usage rates. Estimates suggest a 25% reduction in fatal roadway departure crashes, reducing economic and comprehensive cost by 20.1 billion and 135.5 billion, by the year 2030.

Emergency events can easily cause disruptions or delays in the transportation network, causing performance to decline. Enhancement of the vulnerability of transportation networks contributes to the ability of networks to respond to emergencies. To analyze the multi-layer network’s structure and vulnerability, this paper proposes a transportation network vulnerability assessment model approach taking the highway, high-speed railway, and civil aviation networks in the southwest region of China as case studies. The analysis shows that the vulnerability of the high-speed railway and civil aviation framework could be poor. The highway and high-speed railway framework connect most cities in the two-layer transportation network. However, three-layer transportation networks show better vulnerability. When the disruption proportion reaches 29.74%, the three-layer transportation network collapses into a single-layer or two-layer network. But the network maintains a certain level of operation. This paper provides reference suggestions for the construction of urban traffic terminal expansion. In the southwest region of China, it is necessary to build a three-layer transportation network and accelerate the construction of a synergistic aviation-high speed railway-highway intermodal transportation system.

Autonomous vehicles (AV) have transformed the transportation industry in recent years because of their potential to reduce the number of traffic deaths and injuries, and it is conceivable that when they are integrated with other rideshare networks, they will maximize the utilization of both technologies and become a major mode of public transportation. Safety is the primary concern of major hurdle, and some studies have found that even though AVs have the potential to eliminate driver error, it is unclear whether the magnitude of damage caused by them will be less than that of human-driven conventional vehicles (CV). Since on-demand autonomous vehicles are the future of mobility, and safety remains a major hurdle for their development and commercialization, it is essential that a thorough analysis of the effects of AV-based transportation on traffic crashes be conducted. The goal of this research is to compare the on-road safety of AVs by examining disengagements and actual crashes with CVs. Statistical data from websites like the National Highway Traffic Safety Administration (NHTSA) and the National Safety Council (NSC), as well as data from academic works and research articles, were analyzed, and the results showed AVs have greater potential to reduce injury and fatal crashes compared to CVs; intersections are the hotspots of AV crashes due to the complex traffic environment; and the majority of AV crashes involve rear-end crashes, with CVs hitting the rear of AVs. The findings of this research will aid safety officials in evaluating and enhancing the safety performance of AVs.

Decision-makers in traffic agencies and police departments require a wide variety of high-quality data to support traffic safety problem identification, program implementation, and result evaluation. A major challenge is integrating traffic safety data collected and managed in different databases, often across multiple agencies. This paper describes the design and implementation of a Wisconsin traffic safety data warehouse that is focused on the six core traffic record data systems identified by the National Highway Traffic Safety Administration for its model performance measures. As a first step, the development of the Wisconsin data warehouse is focused on linking crash records to citations and warnings. The design of the data warehouse is determined by the selection of the data source, description of the data flow architecture, and design of the snowflake schema. The snowflake schema was found to support a wide variety of in-depth and flexible traffic safety analyses and is easily extended for future safety applications.

Montgomery County Police Department in Maryland hosted a cannabis intoxication impaired driving lab, to evaluate the driving behavior of medical marijuana users, pre- and post-cannabis consumption. A portable driving simulator with an eye-tracking device was used where 10 participants drove a virtual network in two different events—a traffic light changing from green to yellow and the sudden appearance of a jaywalking pedestrian. An accelerated failure time (AFT) model was used to calculate the speed reduction times in both scenarios. A total of 10 simulation sessions were conducted for 10 participants pre-cannabis consumption and another 10 sessions post-cannabis consumption, which consisted of two doses within a period of 2 h. Two events were evaluated for speed reductions times, a changing traffic light and sudden appearance of a jaywalking pedestrian. It was found that speed reduction times were lower post-cannabis consumption in the event of the traffic light changing from green to yellow. The results showed that the participant speed reduction times are lower; that is, they brake harder post-cannabis consumption, when they encounter a change in traffic light, whereas there is no statistical difference in speed reduction times pre- and post-cannabis consumption, when they encounter a jaywalking pedestrian. The gaze analysis finds no significant difference in eye gaze pre- and post-cannabis consumption in both events.

With the increasing number of road traffic deaths, transportation safety-related study has gained a significant amount of research interest. Tri-axial acceleration values and vehicle speed are observed to be the most efficient vehicular features to identify and measure unsafe driving events. An abnormality index has been proposed to quantify the extent of driving anomaly. The effects of different external stimuli, such as auditory, olfactory, and visual stimuli, over the proposed abnormality index have been studied, and the index is further validated by a set of experimental data collected from a total of 70 sessions of real-time driving experiments. These results can be incorporated further for developing better driving safety systems to cause the advancement of existing autonomous vehicles.

Missing data problem is a major concern in data analysis and statistical modeling. Despite sizable efforts to address the missing data problem in different contexts, the complete-case deletion is still the state of the practice in crash data analysis, which potentially imposes significant bias on subsequent analyses. Besides, most proposed methods are inconsistent with the missing not at random (MNAR) pattern in crash data. In this study, we propose a heuristic optimization algorithm, the genetic algorithm clustered weighting (GCW), to determine a set of weights for the complete records of a data set minimizing the missing bias imposed by the complete-case deletion. We measure the missing bias by the statistical difference of a given set of variables before and after the complete-case deletion. In the proposed algorithm, first, similar complete records are clustered to reduce the computation burden. Subsequently, a genetic optimizer calculates the optimal weights for each cluster. Finally, the optimal weights are assigned to the associated complete records of the data set. We employ five-year crash data containing the MNAR mechanism to appraise the performance of the algorithm. The results demonstrate that the GCW dissipates the missing bias of 95% of the variables that are included in the analysis. On the other hand, in 90% of the excluded variables, the missing bias is significant after implementing the GCW. Moreover, the results of implementing GCW on different data sets suggest that for four out of five data sets, GCW reduces the missing bias by more than 97% compared to the complete-case deletion. The findings accentuate the applicability of the proposed approach for real-sized data and its higher performance over the complete-case deletion.

Research on roundabout safety has shown that roundabouts can significantly reduce fatal and injury crashes but often lead to an increase in property damage only crashes. Most safety evaluations have been conducted using aggregated data from roundabouts with several geometric and operational configurations. Roundabouts may have different safety effects based on traffic control configuration replaced, number of lanes, and location. Limited data has been available to perform safety evaluations for roundabouts with specific configurations. The objective of this study is to evaluate the safety effectiveness of four leg single lane roundabouts replacing two-way stop controlled, all-way stop control, and signalized intersections in rural and urban locations. Sixty-four Wisconsin roundabouts were selected for the evaluation. Geometric (number of lanes and legs), operational (signal control type, speed limit, and average annual daily traffic), and crash data were collected before and after the implementation of the roundabouts. The observational before and after empirical Bayes method was implemented using Wisconsin-specific safety performance functions. The results of the study indicate that single lane roundabouts reduced fatal and injury crashes between 39% and 58% and increased property damage crashes between 33% and 87%. Fatal and injury crashes crash reductions are in line with existing literature; however, property damage crash increases are significantly larger than previously recorded.

Smart work zone (SWZ) systems have received significant attention for improving work zone (WZ) safety and mobility. This pilot study evaluated the effectiveness of Active Work Zone Awareness Devices (AWADs), connected vehicle technologies, law enforcement (LE), and their combinations in improving WZ safety on arterials. Vehicle speed data and driving behavior data were collected at seven sites in Florida that cover various WZ types, roadway classifications, and traffic conditions. Statistical comparisons indicated decreased speeds entering arterial WZs by 10%, increased safe driving behavior by 39%, and reduced risky driving by 44% by deploying AWADs alone. Combining AWAD with the presence of LE increased safe driving behavior by 61% and reduced risky driving by 56%. iCone products that communicate WZ conditions or warnings with road users are anticipated to have an impact on reducing vehicle speeds if the message can be broadcast automatically via navigation systems in the future.

Two-tier traffic control signals (2TCSs) are made up of first traffic control signal (advance amber lights) and second traffic control signal (stop and go lights) spaced at an appropriately predetermined distance apart. The paper investigated the influence of advance amber lights on red light running at signalised intersections where the probability of red light running (RLR) is a measure of 2TCSs’ effectiveness. The paper used observational data because most researchers recognize observational data over simulator data when investigating driver behaviour. Data for the two scenarios (two-tier traffic control signal and traditional traffic control signal) were collected at four signalised intersections in Durban, South Africa. A stochastic model was used to determine the probability of RLR. Results show a significant decrease in the probability of RLR under a two-tier traffic control signals. Based on the synthesis of evidence obtained from the stochastic relationship between red-light running and two-tier traffic control signal, the paper concluded that two-tier traffic control signals have significant positive influence on red light running at signalised intersections. Further, the two-tier traffic control signals are more effective at minimising red light running than the traditional traffic control signals in operation at signalised intersections worldwide.

Adverse weather conditions can be hazardous for driving, especially along high-speed interstate facilities. Despite the significant impact of weather on traffic safety, relatively few studies have documented the effects of real-time weather on crashes. This study investigates the impact of real-time weather-related factors on crash severity along Kentucky’s Interstate 64 (I-64), Interstate 65 (I-65), and Interstate 75 (I-75). Recent five years and four months of crashes (January 1, 2016, through April 30, 2021) along the three aforementioned interstates were collected from the Kentucky Transportation Cabinet. Crashes were merged with real-time weather variables at the time of the crash from the Kentucky Mesonet stations. Five real-time weather-related variables were considered, namely, air temperature (°F), relative humidity (%), precipitation (in.), solar radiation (watts/m²), and wind speed (miles per hour). The three interstates were further categorized into different climate zones based on the geographic location in the state of Kentucky. The severity index “SI” (i.e., ratio of percent severe crashes to percent exposure “or the number of days for a specific weather variable threshold”) was introduced to analyze the weather-related variables. Results from the severity indices showed that crash severity increased with the increase in air temperature, relative humidity, and solar radiation along all three interstates. On the other hand, crash severity decreased with an increase in precipitation (or rainfall), likely due to drivers being more cautious. The severity index for wind speed yielded varying severity impact along the three interstates and even within the climate zone of each interstate. The association rules mining (ARM) technique was also applied to uncover associations between real-time weather variables and severe crash likelihood along Kentucky’s interstates. The findings showed that “northwest of Lexington” climate zone, air temperature, and relative humidity, among others, had significant associations with the severe crash likelihood. The study findings helped to suggest specific weather-related states to feed to dynamic message signs to enhance travelers’ safety along the interstates.

The Dhaka-Sylhet National Highway (N2) of Bangladesh is one of the deadliest highways in South Asia. The main objective of this study was to identify and evaluate the past and present black spots on the first 100 km of this highway, and provide engineering solutions with preliminary cost estimation by using crash and other traffic data analysis, field surveys, and interview of the neighboring people. In order to identify and rank the black spots, detailed historical crash data was first analyzed through six screenings, namely crash number, crash rank, peak searching, crash severity, crash frequency, and KGM method. The older database was compared with the newer database. An assessment of countermeasures provided on N2 in the past is also presented. The research recommended a list of possible countermeasures for each of the black spots. It was found that careless pedestrians, lack of facilities for pedestrians, and careless driving are among the main contributing factors leading to the road crashes on N2.

This research identifies and evaluates the status of road management related to road safety, and provides engineering and other effective solutions for the 8-km-long route from the Kuril Bishwa Road bus stoppage (Pragati Sharani) to Mouchak bus stoppage (DIT Road), specifically for critical intersections, following the International Road Assessment Programme (iRAP) ranking protocol. The government of Bangladesh is interested in developing this route as a standard corridor, because the planned metro rail Line #1 will use this route. The ViDA software and MAAP software were used in this research. The problems identified in this research were based on the accident data collected from the Accident Research Institute (ARI) and the field survey, which are used here as ViDA software’s input along with people’s and traffic management authority’s perceptions. With a target of having a 3-star rating, effective solutions are provided by trial-and-error method, with the help of the ViDA software of iRAP. The research proved that the ViDA software can effectively be used for iRAP ranking of the roads of Dhaka City and similar cities.

Demand-responsive ridesharing services have gained popularity in the recent past and are being adopted by several cities as alternative or means to augment existing public transit services. Empirical studies have explored the impacts of demand-responsive ridesharing services such as congestion, travel patterns, traffic flows, and emissions; however, how these services impact traffic safety in terms of traffic crashes and injuries could not get much attention. Previous studies on safety effects of ridesharing services are based on (un)availability of a ridesharing service; however, this study argues that or absence of a service is not enough to have a significant impact on traffic safety, rather it is that actual usage of a service that makes a difference. This research purposes to investigate the impacts of ridesharing services on traffic safety, using RideAustin, a ridesharing service in Austin, Texas, as a case study. The results show that the number of traffic crashes decreased to statistically significant level with an increase in the number of trips in a block group to a certain threshold. Just presence of the service or very low ridership (less than one ride per block group) did not indicate any effect on the number of crashes indicating no impact on traffic safety. Findings of this analysis will help to understand the safety impacts of ridesharing services.

Pedestrian occlusion is challenging for autonomous vehicles (AVs) at midblock locations on multilane roadways because an AV cannot detect crossing pedestrians that are fully occluded by downstream vehicles in adjacent lanes. This paper tests the capability of vehicle-to-vehicle (V2V) communication between an AV and its downstream vehicles to share midblock pedestrian crossing information. The researchers developed a V2V-based collision-avoidance decision strategy and compared it to a base scenario (i.e., decision strategy without the utilization of V2V). Simulation results showed that for the base scenario, the near-zero time-to-collision (TTC) indicated no time for the AV to take appropriate action and resulted in dramatic braking followed by collisions. But the V2V-based collision-avoidance decision strategy allowed for a proportional braking approach to increase the TTC allowing the pedestrian to cross safely. To conclude, the V2V-based collision-avoidance decision strategy has higher safety benefits for an AV interacting with fully occluded pedestrians at midblock locations on multilane roadways.

With the advancement of the automated vehicle technologies, it is critical to understand the knowledge gap among drivers on the limitations and capabilities of existing advanced driving assistance systems (ADAS) that may contribute to dangerous driving habits and misjudgment. For example, some ADAS include adaptive cruise control (ACC), but many drivers do not know that this feature may not function as expected in response to stationary objects. Variation in terminology, feature capabilities, and advertising have resulted in driver misconceptions and confusion regarding ADAS. This paper reviews these common misconceptions of ADAS capabilities and shortcoming in driver education and behavior. Recommended public messaging was developed to provide simplified, neutral statements for the purpose of driver education.

Connected and automated vehicles (CAVs) have the potential to improve many aspects of the current transportation systems such as safety, mobility, and energy efficiency. In order to evaluate the benefits and impacts of a CAV, the CAV control algorithm is typically implemented on vehicles simulated in a traffic microsimulation environment. However, traffic microsimulation usually lacks detailed vehicle and powertrain dynamics, making it challenging to fully understand how a CAV control algorithm will perform and respond on an actual vehicle. Whether the same benefits measured in the simulation will also be observed in real-world remains an open question. One potential approach to fill in this gap is to conduct a co-simulation of traffic microsimulation with detailed vehicle and powertrain dynamics models, often developed in MATLAB Simulink. However, current microsimulation tools such as VISSIM and SUMO do not have a ready-to-use interface for co-simulation with vehicle dynamics and Simulink. Also, even if such an interface exists, it will be tool-specific, making it challenging to shift from one tool to another or test CAV controls in different tools. There are needs for tool-agnostic co-simulation as different microsimulation tools have their pros and cons, and researchers often need to use different tools based on the purposes of the simulation, project needs, and applications. In this work, Flexible Interface for X-in-the-loop Simulation (FIXS) is developed that can support the co-simulation of microsimulation, CAV control algorithm, and vehicle dynamics model in Simulink. Enabled by the FIXS, the benefit and performance of a CAV control algorithm can be better understood with the consideration of vehicle responses and dynamics. The connection to VISSIM and SUMO is handled internally by the interface, and users can easily switch tools by changing a configuration file. The co-simulation capability is demonstrated for a VISSIM eco-approach and departure CAV scenario and a SUMO cooperative merging scenario for both a passenger CAV and a class 8 heavy-duty connected and automated trucks.

Connected and automated vehicle (CAV) technologies have demonstrated great potential for safety, mobility, environmental, and social benefits. Multiple test beds have been deployed throughout the world to investigate and develop CAV-related applications. Information sharing between CAVs and road infrastructure is receiving significant attention. The Traffic Operations and Safety (TOPS) Laboratory in the UW–Madison College of Engineering has partnered with the city of Madison and the Wisconsin Department of Transportation (WisDOT) to develop a connected corridor as a test bed for CAV and cooperative driving automation (CDA) technologies. In this work, we present an overview of the system architecture, implementation, and operation of the data pipeline system. The system is designed to collect, process, extract, and store real-time data collected and transmitted by the roadside units (RSUs) deployed at the intersections in the connected corridor. A virtual edge computing design is implemented in the data pipeline’s system architecture to enhance the system’s reliability and scalability. The system modules, dataflows, and data visualization are also discussed. The data archived by the data pipeline system will be used for future CAV and CDA applications. The successful implementation of the current prototype is easily transferable and applicable to other connected corridors of a similar nature that will be operated in the future.

Shared autonomous vehicles (SAVs) are among the most anticipated advancements in the smart mobility domain, as they have the potential to revolutionize transportation. The literature includes studies of individuals’ attitudes toward the adoption of SAVs, but it primarily focuses on potential and nonusers of SAVs who have little or no personal experience with the technology. This research concentrates on a pilot project conducted of RAPID (Rideshare Automation and Payment Integration Demonstration), a ridesharing service that operated in Arlington, Texas, from March 2021 to March 2022. It provides a realistic view of how it was received and utilized, and identifies individual attitudes, perceptions, and concerns that arose following its integration into an existing on-demand ridesharing service. A two-step approach was employed for this study. First, it analyzed survey data collected from SAV service users, which revealed that they rated the seating comfort and boarding highly but rated the pickup and drop-off locations very low. Second, a focus group of four individuals was formed of both users and nonusers of the service, and their input was used to formulate in-depth insights into the post-deployment phase of the project. Our findings indicate that service availability, geographic accessibility, and higher cost were the three major concerns of the participants; lack of accessibility and lack of awareness were the two major reasons for not riding the service. This study provides insights into the perceptions and attitudes of SAV users and nonusers that will help transit agencies successfully implement future SAV services and develop and design effective policies.

Shared autonomous vehicles (SAVs) are projected by experts to transform transportation, but consumers will decide whether they are actually a success. This study aims to identify the perceptions, attitudes, concerns, and preferences of people who experienced a pilot demonstration of SAVs that had been integrated with current transportation systems. It applies ordinal logistic regression to analyze the survey data collected from the users and nonusers of the SAV service and analyzes the results of interviews conducted in the post-deployment phase of the project. The results of the survey analysis indicate that the ease of using AVs, which, among other things, eliminates the need for parking, is positively correlated with individuals’ willingness to use SAVs. Conversely, concerns and confusion about having both drivers and AVs on the street are negatively correlated with individuals’ willingness to use SAVs. Virtual interviews with users and nonusers in the post-deployment phase of the project revealed that waiting time, pickup and drop-off locations, and the inability of AVs to make tight turns at intersections were major concerns for SAV service users. This study provides insights into the perceptions and attitudes of those who do and do not use the service and identifies strategies to successfully integrate an SAV service with an existing transportation network to enhance SAV operations.

Driving in foggy conditions poses high risks to road users due to the reduction of visibility, affecting the drivers’ vision and perception, and making changes in driving behavior, which is one of the most important factors affecting vehicular emissions and fuel consumption. This study analyzes the PTV VISSIM traffic microsimulation outputs for exhaust emissions and fuel consumption of vehicles simulated under adverse weather conditions. This weather-dependent simulation is developed by using the advanced psychophysical car-following model “Wiedemann’s 99,” to flexibly control the driving behavior parameters in various driving conditions. Results show that vehicles under foggy conditions consume more fuel and produce more emissions in comparison with clear sky conditions and other scenarios. With the transition of current cities to smart sustainable cities and by introducing automated vehicles (AVs) to the traditional traffic network and gradually increasing their penetration rate, negative environmental impacts of driving under foggy conditions will be reduced, and improvement in overall mobility of a shared network of autonomous and human-driven is observable.

Connected autonomous shared electric vehicles have the potential to reduce operating costs and increase service coverage for TNC. Using an activity-based modeling framework, this research provides a thorough examination of the expanded service coverage of shared mobility brought on by autonomous vehicles (AVs) in lower-density areas. Shared mobility is a cutting-edge transportation strategy that enables users to have momentary access to a mode of transportation as needed. In lower-density areas, shared mobility can improve access, lessen social exclusion, and give carless and transit-dependent households transportation options. The existing transportation network companies (TNC) provide limited service in low-density areas for lesser profits than in urban areas causing greater wait times. Very few studies have explored the impact of improved TNC coverage in low-density areas. With the usage of autonomous vehicles in TNC, people in low-density areas will be more attracted to shared mobility with lower cost and wait time. The study used an activity-based modeling framework in the Southeast Florida Regional Planning Model (SERPM) to estimate the effect on the transportation system by changing service coverage in low and very low-density areas. The model run from different scenarios demonstrated that with the increased service coverage the TNC membership increases for the household and auto ownership declines. However, the vehicle mile traveled (VMT) and vehicle hour traveled (VHT) increased for having zero passenger trips. The study will guide transportation planning organizations to assess the behavioral change in low-density areas for autonomous vehicles’ presence in TNC.

Highway safety is still a concerning issue despite different safety strategies, improvements in infrastructure, and in-vehicle technologies. Connected and automated vehicle (CAV) technology offers the possibility of improving highway safety through data exchange. Using a driving simulator and an eye-tracking system, this study investigates drivers’ behavior while influenced by certain CAV functions near a work zone. The variables obtained from the driving simulator computer include driving speed, acceleration/deceleration, and takeover reaction time (ToRt) when approaching work zone and driving through a work zone with and without CAV capabilities are compared. The eye-tracking heat map generated with gaze data acquired from eye-tracking glass proves that all participants noticed the takeover request (TOR) and responded accordingly. The analysis of variance (ANOVA) results indicate that the driving performance in two scenarios is different within the work zone and a buffer zone segment. The jerk analysis results suggest that the participants tend to drive with a lower frequency of acceleration and deceleration rate in the base scenario that has no autonomous feature in comparison to CAV scenario. Regression analysis shows that participants who drive more mileages annually have a shorter ToRt. Interestingly, ToRt decreases by age, and participants who trust CAV features have longer ToRt. The overall results suggest that CAV feature is capable of completing the primary intention of reducing speed in the vicinity of work zone to improve safety.

With the advent of advanced driver assistance systems (ADAS), there is an increasing need to evaluate driver behavior while using such technology. In this unique study, a forward collision warning (FCW) system using connected vehicle technology was introduced in a driving simulator environment, to evaluate driver braking behavior, and then the results are validated using data from field tests. A total of 93 participants were recruited for this study, for which a virtual network of South Baltimore was created. A one sample t-test was conducted, and it was found that the mean reduction in speed of 15.07 mph post-FCW is statistically significant. A random forest machine learning algorithm was found to be the best fit for ranking the most important variables in the data set by order of importance. Field data obtained from the University of Michigan Transportation Research Institute (UMTRI) substantiated the FCW findings from this driving simulator study. Our real-world findings confirmed the conclusions drawn from the simulation experiment as it was also found that the presence of the FCW had a positive and significant impact on the change of speed, which proves the effectiveness of our simulation experiment. Moreover, this field experiment found that the majority of drivers started pressing the brake pedal after receiving the warning, which shows the effectiveness of the FCW system. Thus, from both the simulator and the field tests, it can be said that an FCW system is effective in inducing a speed reduction.

With the advent of connected and automated vehicle (CAV) technology, there is an increasing need to evaluate driver behavior while using such technology. In this first-of-a-kind study, a pedestrian collision warning (PCW) system, using CAV technology, was introduced in a driving simulator environment, to evaluate driver braking behavior, in the presence of a jaywalking pedestrian. A total of 93 participants from diverse socioeconomic backgrounds were recruited for this study, for which a virtual network of downtown Baltimore was created. An eye-tracking device was also used to observe distractions and head movements. A log logistic accelerated failure time (AFT) distribution model was used for this analysis, to calculate speed reduction times: time from the moment the pedestrian becomes visible, to the point where a minimum speed was reached, to allow the pedestrian to pass. The presence of the PCW system significantly impacted the speed reduction time and deceleration rate, as it increased the former and reduced the latter, which proves the effectiveness of this system in providing an effective driving maneuver, by drastically reducing speed. A jerk analysis is conducted to analyze the suddenness of braking and acceleration. Gaze analysis showed that the system was able to attract the attention of the drivers, as the majority of the drivers noticed the displayed warning. The familiarity of the driver with the route and connected vehicles reduces the speed reduction time; gender also can have a significant impact as males tend to have longer speed reduction time, that is, more time to comfortably brake and allow the pedestrian to pass.

Driver behavior at rural horizontal curves has long been a topic of interest from safety and operational standpoints. This interest stems from the tendency for these sites to incur substantial severe vehicle crashes. Driver error typically occurs at locations where inconsistent design of the roadway alignment fails to meet driver expectancies (e.g., sharp curve following a straight tangent). Numerous studies have developed speed choice models to understand this trend. The emergence of autonomous driving systems has led a renewed interest to develop and implement robust steering choice models for navigating curves. This paper presents a framework for simulating human-driven vehicles, autonomous vehicles (AVs), and connected autonomous vehicles (CAVs) that considers drivers’ speed and steering choices in response to roadway alignment. The results of the analysis indicate that speed, lateral lookahead distance, and safety performance are interwoven. Improper selection of either will cause discomfort for passengers at best and contribute to oversteering into the oncoming lane at worst.

The potential impact of autonomous vehicles (AVs) on traffic citation revenues has been widely discussed. While not many vehicles in the US meet the requirements to be truly autonomous, a vast majority of newer vehicles contain several advanced driver assistance systems (ADASs). Many of these ADASs can improve both driver behavior and safety. Thus, these ADASs can affect traffic citations and the local/state revenues. In this study, we qualitatively identify ADASs that have potential to impact citations, and we qualitatively identify citations that have the most potential to be reduced by the introduction of various ADASs. We concluded that many of the ADASs that currently exist can prevent or reduce the frequency of common citation types. While a reduction in citations might sound attractive to many individuals, a reduction in revenue largely impacts local and state governments and can ultimately affect these individuals.

Recent technological advancements, especially automated vehicles (AVs), are likely to reduce human involvement while driving by assisting drivers in various tasks such as steering, accelerating, or braking. The effect of AVs with different levels of automation on crash occurrence, especially considering fatal crashes, could vary based on road geometry and weather conditions. In addition, the factors affecting crashes where AVs are at-fault could differ from those affecting crashes where AVs are not at-fault, demanding further research. Therefore, this study focuses on identifying the factors affecting fatal crashes involving at-fault and not-at-fault level 1 and level 2 AVs using crash data from the United States. The partial proportionality odds modeling technique was used to develop two separate models for crashes involving at-fault and not-at-fault AVs. The results showed that level 1 and level 2 AVs have higher odds of getting involved in crashes with pedestrians and bicyclists due to their fault. In addition, they have higher odds of getting involved in crashes at intersections, at entry/exit ramps, and on one-way roads due to the fault of other drivers. The results from this study help enhance traffic safety, considering crash involvement factors of not-at-fault and at-fault AVs. Moreover, they assist the practitioners in developing safety countermeasures and manufacturers to improve smart features, especially for AVs with higher levels of automation.

Although many studies have suggested that connected and autonomous vehicle (CAV) technologies can improve intersection operation, few literature studies reveal the effect of timely identified traffic characteristics at signalized intersections with the CAV-generated data on adaptive signal control schemes. This paper presents a methodology to transform the emerging data source into useful real-time information and associated parameters that can infer traffic states and be used for innovating signal control. Two scenarios are designed at an intersection in Cincinnati, Ohio. Scenario 1 simulates a case as arrival patterns vary from 100 to 2,000 vehicles/h at an increment of 100 vehicles/h on each approach. Scenario 2 simulates a case as the volume ratio varies from 0.1 to 0.9 at an increment of 0.1 each time. The results indicate an up to 84% of increased intersection throughput with the adaptive signal plan by using the CAV-generated data, compared with the pretimed signal plan.

Realistic and accurate evaluations of the potential influences of autonomous vehicles (AVs) on transportation systems are based on an adequate understanding of the market penetration and customers’ preferences for various AV technologies and services. Although research efforts have been made to enhance public knowledge about what population segments could be early AV passengers, few attempts have been made to examine the change in public attitudes toward AVs over time. Previously, surveys were the primary approach to measure public acceptance, preferences, and contributing factors. However, both the measures of approval and their corresponding explanatory variables vary substantially across research articles. This paper explores the changes in public perceptions toward AV technologies over time. A synthesis of previously published survey results related to AV technologies was conducted. This study may help provide insights into how public perceptions toward AVs have changed over time and the components that change public perceptions.

This work is based on a project to develop a physics-based, 3D digital visual environment that is a replication of actual field conditions for over 70 mi of Ohio highways and city roads for use in a driving simulator for the Ohio Department of Transportation. While transportation engineering design traditionally involves 3D design in a 2D workspace to create the built environment in the context of a natural environment, this project required replication of existing natural + built environments in a 3D digital space, thereby presenting a unique challenge to develop a new, repeatable process to create a specific digital end product. Using industry-specific software comprised of InfraWorks (urban infrastructure design), Civil 3D (terrain modeling), Rhino (3D product modeling), 3ds Max (rendering/animation), Maya (3D animation/simulation), and Python (scripting) that are traditionally dedicated to their fields, the team developed a process to integrate them outside of their intended purposes so that they could connect industry-specific functionalities to deliver a novel product that can now be utilized by multiple markets. This process utilizes the functionalities of each software to resolve a portion of the puzzle and delivers it as a solution for the next step of development using another software. Using an iterative development cycle approach, the process bridges the gaps between the industries of Transportation Engineering, Visualization, Architecture, and Gaming to deliver the end product. The resulting 3D digital model of the existing urban environment can now be readily used as a baseline product for any industry that would benefit from such a digital model. In transportation engineering, it can be used in Transportation Systems Planning, Surface Operations, and/or Workforce Development. In outside/connected markets, it can be used in UI-based development, interactive game-based multiplayer virtual meetings, and photo-realistic immersive models for use in VR/multiplayer exploratory environments. This process has been standardized for the digital development of existing site conditions and context for architectural conceptualization of buildings and public spaces in the Architecture program at the University of Cincinnati. The same process has been carried into the next phase of development for the Ohio Department of Transportation.

Over the decades, significant efforts have been made in evaluating pavement condition by mounting cameras on a vehicle and automatically analyzing images by incorporating digital image processing techniques. Despite recent advances, there are still some limitations associated with current automated crack collection and analysis systems such as potential risk to vehicular safety, limited coverage by cameras, traffic blocking camera views, and errors in analyzing images for crack extent and severity. Inspired by current developments of deep learning technology such as object classification and sematic segmentation along with unmanned aerial vehicle (UAV) technology, this paper presents a set of comprehensive automated crack analysis algorithms based on a combination of deep learning and UAV images. To extract pavements from UAV images and segment cracks from extracted pavement images, contracting encoder path of the U-Net model was modified with various deep learning models. A tile-based pavement crack analysis system was then developed to measure percent cracking and crack widths from segmented crack images. It can be concluded that the developed pavement extraction and crack analysis system using UAV images will help public agencies evaluate pavement conditions in a systematic and cost-effective manner.

A digital twin involves a realistic digital model, sensing of real-world parameters, depiction of those parameters in a digital environment, and simulation of the virtual system. While various methods exist to create digital models of geographic locations and their road networks, little work has been done to automatically track vehicles from video and generate virtual vehicles in the digital model. This paper presents one pipeline for performing vehicle detection and tracking on an edge device, and the results are visualized in a university campus digital model running on a workstation. Challenges stemming from camera lens effects, irregularities in the digital model, and object tracking uncertainties are discussed. The precise transformation of 2D vehicle detection to 3D representation is the main challenge addressed in this work. The approach presented focuses on calibrating different parts of the scene separately. The focus of future work will be to make this calibration process semi-automated.

Travel time prediction plays a significant role in traffic data analysis. In the area of travel time prediction, traditional methods such as linear regression and time series models have been widely applied to predict travel times using historical travel time data. However, with the development of artificial intelligence technologies, these traditional models may become replaceable. In recent years, various novel machine learning-based data prediction methods have been developed accordingly. These types of approaches aim to increase the efficiency and accuracy of the prediction. This study intends to employ ensemble learning-based models (including the XGBoost model and RF model) to predict freeway travel time using information such as time of day, day of the week, and weather conditions. The models are validated on a freeway corridor in the area of Charlotte, North Carolina. The temporal and spatial correlations between each segment are also considered in the model. The effects of different feature combinations in the model are investigated and quantified. The modelling results indicate that the XGBoost model has considerable advantages in terms of both prediction accuracy and efficiency. The research findings demonstrate the feasibility of applying XGBoost in travel-time prediction and prove that the XGBoost model is applicable and performs well.

As the connected vehicle (CV) technology is being transformed into transportation systems, it is an emerging need to test the effect of the variable speed limit (VSL) strategy on CVs. To better understand the VSL effects under varied weather conditions, we propose to classify the weather intensity with the developed genetic fuzzy logic-based (GFLB) VSL control model for adjusting the speed limit versus various weather conditions. In this model, a genetic algorithm (GA) is integrated with both safety and speed-related operation efficiency into the fitness function into the intelligent driver model (IDM), which is calibrated with the real-time car-following behaviors. The simulation-based tests of the GFLB VSL model using the data on the I-80 corridor, Wyoming, indicate that the VSL outperforms the non-speed-limit-control and fixed-speed-limit-control strategies under sunny, foggy, and snowy weather conditions, given the same weight to safety and mobility variables in the modeling objective function.

Considering recent advances in remote sensing satellite systems and computer vision algorithms, many satellite sensing platforms and sensors have been used to monitor the condition and usage of transportation infrastructure systems. The level of details that can be detected increases significantly with the increase of ground sample distance (GSD), which is around 15 cm–30 cm for high-resolution satellite images. In this study, we analyzed data acquired from high-resolution satellite imagery to provide insights, predictive signals, and trend for travel demand estimation. More specifically, we estimate the impact of COVID-19 in the metropolitan area of Houston using satellite imagery from Google Earth Engine data sets. We developed a car-counting model through Detectron2 and Faster R-CNN to monitor the presence of cars within different locations (i.e., universities, shopping malls, community plazas, restaurants, and supermarkets) before and during the COVID-19. The results show that the number of cars detected at these selected locations reduced on average 30% in 2020 compared with the previous year 2019. The results also show that satellite imagery provides rich information for travel demand and economic activity estimation. Together with advanced computer vision and deep learning algorithms, it can generate reliable and accurate information for transportation agency decision makers.

Due to the high demand for vehicles in the increasing middle class, the massive crap tires have been an environmental issue. Ground tire rubber (GTR), that is produced from crap tires and used in pavement construction, has proven to be a primary method to get rid of the piles of tire landfills. Although asphalt pavement using dry process demonstrates less popularity than wet process, some recent studies have shown successful experience with dry process GTR mixes. To promote the successful use of GTR in Oklahoma, the Oklahoma Department of Transportation (ODOT) constructed a GTR test section on State Highway 33 (SH-33) for further identifying the benefit of dry process GTR. From March 2019 to July 2021, six field data collections were performed at intervals of pre-placement, post-placement, three months, nine months, one year, and two years. In each data collection, pavement cracking and rutting were evaluated via the Pave3D 8K system, pavement macro-texture and roughness were assessed via the Ames profiler, and pavement friction was collected via the grip tester. It is demonstrated that the GTR site exhibits comparable performance to that of the abutting warm mix asphalt (WMA) section in terms of rutting, roughness, texture, and friction, while with lower field cracking resistance.

Accurately predicting travel time information can be helpful for travelers. This study proposes a framework for predicting network-level travel time index (TTI) using machine learning models. A case study was performed on more than 50,000 TTI data collected from the Washington, DC, area over six years. The proposed approach is also able to identify the effects of weather and seasonality. The performances of the machine learning models were assessed and compared with each other. It was shown that the ridge regression model outperformed the other models in both short-term and long-term predictions.

Identification of factors affecting the vehicle following-distance behavior in the context of improving safety and reducing crashes is of utmost importance. However, earlier research on the factors influencing following-distance behavior has primarily relied on correlation/association-based techniques, which miss the underlying causal relations between the influencing factors and are therefore unable to identify the actual root causes. Therefore, the aim of this study is to create a causal model by using a framework for causal inference to identify the factors that contribute to participants’ following-distance behavior and to quantitatively analyze their impacts. To realize this aim, we performed human-subject experiments in immersive virtual environments (IVE) integrated with a driving simulator to elicit participants’ following-distance behavior data. IVEs provide users with a fully immersive simulation experience and allow researchers to easily design and manipulate contextual factors and observe the outcomes. As such, incorporating IVEs with causal inference into the following-distance behavior analysis helps to advance a more customizable research tool for investigating causal relationships in safety and traffic capacity studies. Six experimental scenarios involving several road-related factors and human-related factors were performed with 24 participants. Using the PC-stable algorithm and the elicitation method, a causal model was learnt from the data that visualized the causal relationships between the factors. The findings of this analysis suggest that the traffic conditions (i.e., normal, medium, and heavy traffic) have a causal effect on the participants’ following-distance behavior, thus providing an extension to the previous studies where only correlations/associations were observed between those factors.

With an increase in multimodal fatalities and data-driven transportation funding, state agencies are required to justify their investments with robust quantitative data and measures. Thus, there is a critical need for accurate, reliable, and comprehensive information about multimodal travel movements to support the planning, design, and management of smart infrastructure as part of a larger regional surface transportation system. Irrespective of the application, collecting observational data for multiple modes is considered a challenging task due to the unorganized nature of multimodal traffic compared to vehicular traffic. In this study, the research team identifies the state-of-the-art and the state-of-the-practice multimodal data collection techniques used for safety analysis. Based on that, the research team conducts a nationwide outreach effort consisting of interviews with subject matter experts and Metropolitan Planning Organizations and Departments of Transportation employees. This paper summarizes the study development methodology, as well as questions and lessons learned from one-on-one interviews. Additionally, it includes a summary and conclusions regarding video analytics and nationwide best practices of multimodal data collection, analytics, and applications to safety and operations. The outcomes of this study and interviews aim to guide decision makers including the city of Austin regarding the deployment of detection technologies.

Truck parking along the Mid-America Association of State Transportation (MAASTO) region’s major freight corridors is complicated by an overall deficit in needed parking spaces, delivery staging at major logistic centers or urban areas, and hour-of-service (HOS) compliance. These constraints lead to unauthorized or potentially unauthorized parking. To improve freight operations’ safety and offer better parking information, the research team and MAASTO collaborated to build the multi-state Truck Parking Information Management System (TPIMS) and the data archival system (TPIMS-DAS). The TPIMS provides real-time parking availability information to truck drivers, dispatchers, and traffic management centers (TMCs). Truck operators’ use of this information provides for better planning and decision-making and less wasted search time in securing truck parking. This paper presents the development of the TPIMS web portal and practical user management experiences in the development of the data management and provision system. The web portal supports project and research data needs through truck parking data downloads, user account management, and performance measure report downloads. The user management tool on the web portal offers a tiered user data access privilege to meet the system administration needs of MAASTO members. We also present findings from analyzing the parking utilization data archived by TPIMS-DAS over the last three years. These findings elucidate truck drivers’ work patterns, site utilization patterns in different time intervals, and other patterns, such as holiday trends and regional relationships. Information provided by this data set and this analysis can provide guidance for truck parking management and enhancements.

As highway projects become more complex, Departments of Transportation (DOTs) are pressured to complete projects with tighter schedules, stricter budgets, and limited workforce. This motivated DOTs to utilize technology, notably wireless and sensing ones, that helped projects report beneficial outcomes like improved resource utilization and enhanced quality management. This study focuses on 10 technologies—radio frequency identification, barcodes and readers, object recognition, infrared, e-ticketing, ground penetrating radar, unmanned aerial systems or vehicles, light detection and ranging, geographic information system, and global positioning system. A survey was sent out to DOT construction and asset management practitioners to examine the current state of the technologies’ usage, their maturity level, challenges that DOTs faced during implementation, and the key performance indicators that DOTs use to evaluate their performance. Findings provide both practitioners and researchers with a fundamental overview of the current state of the adoption of wireless and sensing technologies in highway projects.

According to the definition given by the United Nations, the aging society is a society of which the proportion of 65-year-old and above is over 7%. However, this proportion in Taiwan has already reached 14% in 2018 and is even estimated to exceed 20% by the year of 2025, leading Taiwan into a super-aged society. The dependency on motorcycle, which is the dominant transportation mode for Taiwanese, is one of the key factors affecting the use of public transportation. Comparing to the cheap and easy-to-park scooters, city bus with very few frequencies and longer journey time is often the least preferred mode of travel for Tainan and Kaohsiung commuters. As a result, the captive users for bus are usually the elderly people, students, and new immigrant workers. Due to the unpopularity of city bus, many elder people in Taiwan became highly dependent users for scooters. Consequently, accident rate for elderly people is much higher compared to other user groups. To improve the service quality and to reduce the labor costs for city bus operations, Kaohsiung and Tainan recently introduced pilot projects regarding autonomous minibus services to communities in the rural townships and the neighborhoods nearby high-speed rail stations. The aims for these projects are to provide an efficient and safe public transportation to meet the needs for transfer trips and nursing care in these communities. Our study seeks to learn about the acceptance for driverless minibus among elderly people. Data regarding travel habits and modal choice preferences were collected from survey questionnaires through our targeted research groups in Tainan, Kaohsiung, and Taipei. And a stated preference demand model was calibrated using the data. Next, we estimate the potential demand for self-driving minibus services and explore the main factors affecting users’ preference. In conclusion, we find that in some scenarios, the driverless DRTS with on-board assistants will be even more popular than scooters and become the most preferred mode of travel. On the other hand, if on-board assistant was absent, then the driverless DRTS will lose more than half of the potential elderly users. The average proportion for the elderly to take driverless DRTS is about 36% from all four scenarios, making driverless DRTS to be the second most popular mode followed by the first-place scooter. Meanwhile, the elderly people in Kaohsiung and Tainan are more likely to take driverless DRTS compared to the elderly people in Taipei. One of the main reasons is that Taipei has much better service quality for rapid transit and bus compared to Tainan and Kaohsiung.

Recent advancements in information and communication technology have contributed to the rise of a new transportation concept: on-demand ridesourcing that uses a smartphone application to connect customers and drivers for real-time and demand-responsive trips. Due to its rapid growth, the ridesharing market is expected to eventually attract a large number of new companies, and empirical studies are focusing on identifying the factors that affect customer satisfaction. There is limited knowledge, however, about the factors that affect students’ satisfaction with ridesharing services, and this study aims to fill that gap by uncovering the significant factors affecting students’ satisfaction levels with conventional fixed routes and on-demand ridesharing services. Data collected from a survey distributed among university students was used as the basis for this study, and the results indicated that users are satisfied with the services overall, but unsatisfied with their waiting times and availability. The outcomes of this study will assist ridesharing companies, transit agencies, and policymakers by providing crucial insights into the factors impacting user satisfaction, which will help them develop policies to enhance their operations and increase the number of individuals availing themselves of the services they offer.