Safety issues at school zone areas have been one of the most important topics in the traffic safety field. Although many studies have evaluated the effectiveness of various traffic control devices (e.g., sign, flashing beacon, speed monitoring display), there is a lack of studies exploring different roadway countermeasures and the relationship between school-related factors and crashes. In this study, the most crash-prone school zone was identified in Orange and Seminole Counties, Florida, based on crash rate. Afterward, a microsimulation network was built in VISSIM environment to test different roadway countermeasures in the school zones. Three different countermeasures: two-step speed reduction (TSR), decreasing the number of driveways (DD), and replacing the two-way left-turn lane (TWLTL) to the raised median (RM) were implemented in the microsimulation. Three surrogate safety measures-: (1) time exposed time to collision (TET), (2) time integrated time to collision (TIT) and (3) time exposed rear-end crash risk index (TERCRI) were utilized in this study as indicators for safety evaluation. The higher value of surrogate safety measures indicates higher crash risk. The results showed that both TSR and DD reduced TET, TIT and TERCRI values significantly compare to the base condition. Moreover, the combination of TSR and DD countermeasures outperformed their individual effectiveness. The One-way ANOVA analysis showed that all the sub-scenarios were significantly different from each other. Sensitivity analysis result has proved that all the sub-scenarios in TSR and DD reduced TET, TIT and TERCRI values significantly for different value of TTC threshold. On the other hand, for converting the TWLTL to RM, the crash risk was higher than the base condition because of the turning movements of vehicle. The results of this study could help transportation planners and decision makers to understand the effect of these countermeasures to improve safety at school zones.

This thesis presents an automated traffic safety diagnostics solution using deep learning techniques to process traffic videos by Unmanned Aerial Vehicle (UAV). Mask R-CNN is employed to better detect vehicles in UAV videos after video stabilization. The vehicle trajectories are generated when tracking the detected vehicle by Channel and Spatial Reliability Tracking (CSRT) algorithm. During the detection process, missing vehicles could be tracked by the process of identifying stopped vehicles and comparing Intersect of Union (IOU) between the tracking results and the detection results. In addition, rotated bounding rectangles based on the pixel-to- pixel manner masks that are generated by Mask R-CNN detection, which are also introduced to obtain precise vehicle size and location data. Moreover, surrogate safety measures (i.e. post- encroachment time (PET)) are calculated for each conflict event at the pixel level. Therefore, conflicts could be identified through the process of comparing the PET values and the threshold. To be more specific, conflict types that include rear-end, head-on, sideswipe, and angle could be determined. A case study is presented at a typical signalized intersection, the results indicate that the proposed framework could notably improve the accuracy of the output data. Furthermore, by calculating the PET values for each conflict event, an automated traffic safety diagnostic for the studied intersection could be conducted. According to the research, rear-end conflicts are the most prevalent conflict type at the studied location, while one angle collision conflict is identified at the study duration. It is expected that the proposed method could help diagnose the safety problems efficiently with UAVs and appropriate countermeasures could be proposed after then.

Connected and automated vehicle (CAV) technologies have recently drawn an increasing attention from governments, vehicle manufacturers, and researchers. Connected vehicle (CV) technologies provide real-time information about the surrounding traffic condition (i.e., position, speed, acceleration) and the traffic management center's decisions. The CV technologies improve the safety by increasing driver situational awareness and reducing crashes through vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I). Vehicle platooning with CV technologies is another key element of the future transportation systems which helps to simultaneously enhance traffic operations and safety. CV technologies can also further increase the efficiency and reliability of automated vehicles (AV) by collecting real-time traffic information through V2V and V2I. However, the market penetration rate (MPR) of CAVs and the higher level of automation might not be fully available in the foreseeable future. Hence, it is worthwhile to study the safety benefits of CAV technologies under different MPRs and lower level of automation. None of the studies focused on both traffic safety and operational benefits for these technologies including different roadway, traffic, and weather conditions. In this study, the effectiveness of CAV technologies (i.e., CV /AV/CAV/CV platooning) were evaluated in different roadway, traffic, and weather conditions. To be more specific, the impact of CVs in reduced visibility condition, longitudinal safety evaluation of CV platooning in the managed lane, lower level of AVs in arterial roadway, and the optimal MPRs of CAVs for both peak and off-peak period are analyzed using simulation techniques. Currently, CAV fleet data are not easily obtainable which is one of the primary reasons to deploy the simulation techniques in this study to evaluate the impacts of CAVs in the roadway. The car following, lane changing, and the platooning behavior of the CAV technologies were modeled in the C++ programming language by considering realistic car following and lane changing models in PTV VISSIM. Surrogate safety assessment techniques were considered to evaluate the safety effectiveness of these CAV technologies, while the average travel time, average speed, and average delay were evaluated as traffic operational measures. Several statistical tests (i.e., Two sample t-test, ANOVA) and the modelling techniques (Tobit, Negative binomial, and Logistic regression) were conducted to evaluate the CAV effectiveness with different MPRs over the baseline scenario. The statistical tests and modeling results suggested that the higher the MPR of CAVs implemented, the higher were the safety and mobility benefits achieved for different roadways (i.e., freeway, expressway, arterials, managed lane), weather (i.e., clear, foggy), and traffic conditions (i.e., peak and off-peak period). Interestingly, from the safety and operation perspective, at least 30% and 20% MPR were needed to achieve both the safety and operational benefits of peak and off-peak period, respectively. This dissertation has major implications for improving transportation infrastructure by recommending optimal MPR of CAVs to achieve balanced mobility and safety benefits considering varying roadway, traffic, and weather condition.

Managed lanes (MLs) have been implemented as a vital strategy for traffic management and traffic safety improvement. The majority of previous studies involving MLs have explored a limited scope of the impact of the MLs segments as a whole, without considering the safety and operational effects of the access design. Also, there are limited studies that investigated the effect of connected vehicles (CVs) on managed lanes. Hence, this study has two main objectives: (1) the first objective is achieved by determining the optimal managed lanes access design, including accessibility level and weaving distance for an at-grade access design. (2) the second objective is to study the effects of applying CVs and CV lanes on the MLs network. Several scenarios were tested using microscopic traffic simulation to determine the optimal access design while taking into consideration accessibility levels and weaving lengths. Both safety (e.g., standard deviation of speed, time-to-collision, and conflict rate) and operational (e.g., level of service, average speed, average delay) performance measures were included in the analyses. For the first objective, the results suggested that one accessibility level is the optimal option for the 9-mile network. A weaving length between 1,000 feet to 1,400 feet per lane change was suggested based on the safety analysis. From the operational perspective, a weaving length between 1,000 feet and 2,000 feet per lane change was recommended. The findings also suggested that MPR% between 10% and 30% was recommended when the CVs are only allowed in MLs. When increasing the number of MLs, the MPR% could be improved to reach 70%. Lastly, the findings proposed that MPR% of 100% could be achieved by allowing the CVs to use all the lanes in the network.

Pedestrian-related crashes are a significant safety issue in the United States and cause considerable amounts of deaths and economic cost. Pedestrian safety is an issue that must be uniquely evaluated in a college campus, where pedestrian volumes are dense. The objective of this research is to identify issues at specific locations around UCF and suggest solutions for improvement. To address this problem, a survey that identifies pedestrian safety issues and locations is distributed to UCF students and staff, and an evaluation of drivers reactions to pedestrian to vehicle (P2V) warning systems is studied through the use of a NADS MiniSim driving simulator. The survey asks participants to identify problem intersections around campus and other issues as pedestrians or bicyclists in the UCF area. Univariate probit models were created from the survey data to identify which factors contribute to pedestrian safety issues, based off the pedestrian's POV and the driver's POV. The models indicated that the more one is exposed to traffic via walking, biking, and driving to campus contributes to less safe experiences. The models also show that higher concerns with drivers not yielding, unsafety of crossing the intersections, and the number of locations to cross, indicate less safe pedestrian experiences from the point of view of pedestrians and drivers. A promising solution for pedestrian safety is Pedestrian to Vehicle (P2V) communication. This study simulates P2V connectivity using a NADS MiniSim Driving Simulator to study the effectiveness of the warning system on drivers. According to the results, the P2V warning system significantly reduced the number of crashes in the tested pre-crash scenarios by 88%. Particularly, the P2V warning system can help decrease the driver's reaction time as well as impact velocity if the crash were to occur.

The thesis contributes to our understanding of incorporating heterogeneity in discrete choice models with respect to exogenous variables and decision rules. Specifically, we evaluate latent segmentation based mixed models that allow for segmenting population based on decision rules while also incorporating unobserved heterogeneity within the segment level decision rule models. In our analysis, we choose to consider the random utility framework along with random regret minimization approach. Further, instead of assuming the number of segments (as 2), we conduct an exhaustive exploration with multiple segments across the two decision rules. Within each segment we also allow for unobserved heterogeneity. The model estimation is conducted using a stated preference data from 695 commuter cyclists compiled through a web-based survey. The probabilistic allocation of respondents to different segments indicates that female commuter cyclists are more utility oriented, however the majority of the commuter cyclist's choice pattern is consistent with regret minimization mechanism. Overall, cyclists' route choice decisions are influenced by roadway attributes, cycling infrastructure availability, pollution exposure, and travel time. The analysis approach also allows us to investigate time based trade-offs across cyclists of different classes. Interestingly, we observed that the trade-off values in regret and utility based segments for roadway attributes are similar in magnitude; but the values differ greatly for cycling infrastructure and exposure attributes, particularly for maximum exposure levels.

Deep learning coupled with existing sensors based multiresolution traffic data and future connected technologies has immense potential to improve traffic operation and management. But to deal with complex transportation problems, we need efficient modeling frameworks for deep learning models. In this study, we propose two different modeling frameworks using Deep Long Short-Term Memory Neural Network (LSTM NN) model to predict future traffic state (speed and signal queue length). In our first problem, we present a modeling framework using deep LSTM NN model to predict traffic speeds in freeways during regular traffic condition as well as under extreme traffic demand, such as a hurricane evacuation. The approach is tested using real-world traffic data collected during hurricane Irma's evacuation for the interstate 75 (I-75), a major evacuation route in Florida. We perform several experiments for predicting speeds for 5 min, 10 min, and 15 min ahead of current time. The results are compared against other traditional prediction models such as K-Nearest Neighbor, Analytic Neural Network (ANN), Auto-Regressive Integrated Moving Average (ARIMA). We find that LSTM-NN performs better than these parametric and non-parametric models. Apart from the improvement in traffic operation, the proposed method can be integrated with evacuation traffic management systems for a better evacuation operation. In our second problem, we develop a data-driven real-time queue length prediction technique using deep LSTM NN model. We consider a connected corridor where information from vehicle detectors (located at the intersection) will be shared to consecutive intersections. We assume that the queue length of an intersection in the next cycle will depend on the queue length of the target and two upstream intersections in the current cycle. We use InSync Adaptive Traffic Control System (ATCS) data to train a Long Short-Term Memory Neural Network model capturing time-dependent patterns of a queue of a signal. To select the best combination of hyperparameters, we use sequential model-based optimization (SMBO) technique. Our experiment results show that the proposed modeling framework performs very well to predict the queue length. Although we run our experiments predicting the queue length for a single movement, the proposed method can be applied for other movements as well. Queue length prediction is a crucial part of an ATCS to optimize control parameters and this method can improve the existing signal optimization technique for ATCS.

Transportation sector is one of the largest emission sources and is a cause for human health concern due to the high dependency on personal vehicle in the U.S. Transportation mode choice studies are currently limited to micro- and regional-level boundaries, lacking of presenting a complete picture of the issues, and the root causes associated with urban passenger transportation choices in the U.S. Hence, system dynamics modeling approach is utilized to capture complex causal relationships among the critical system parameters affecting alternative transportation mode choices in the U.S. as well as to identify possible policy areas to improve alternative transportation mode choice rates for future years up to 2050. Considering the high degree of uncertainties inherent to the problem, multivariate sensitivity analysis is utilized to explore the effectiveness of existing and possible policy implications in dynamic model in the terms of their potential to increase transit ridership and locating critical parameters that influences the most on mode choice and emission rates. Finally, the dissertation advances the current body of knowledge by integrating discrete event simulation (multinomial fractional split model) and system dynamics for hybrid urban commuter transportation simulation to test new scenarios such as autonomous vehicle (AV) adoption along with traditional policy scenarios such as limiting lane-mile increase on roadways and introducing carbon tax policy on vehicle owners. Overall, the developed simulation models clearly indicate the importance of urban structures to secure the future of alternative transportation modes in the U.S. as the prevailing policy practices fail to change system behavior. Thus, transportation system needs a paradigm shift to radically change current impacts and the market penetration of AVs can be one of the reforms to provoke this transition since it is expected to revolutionize mode choice, emission trends, and the built environment.

In the context of pro-active traffic management, real-time safety evaluation is one of the most important components. Previous studies on real-time safety analysis mainly focused on freeways, seldom on arterials. With the advancement of sensing technologies and smart city initiative, more and more real-time traffic data sources are available on arterials, which enables us to evaluate the real-time crash risk on arterials. However, there exist substantial differences between arterials and freeways in terms of traffic flow characteristics, data availability, and even crash mechanism. Therefore, this study aims to deeply evaluate the real-time crash risk on arterials from multiple aspects by integrating all kinds of available data sources. First, Bayesian conditional logistic models (BCL) were developed to examine the relationship between crash occurrence on arterial segments and real-time traffic and signal timing characteristics by incorporating the Bluetooth, adaptive signal control, and weather data, which were extracted from four urban arterials in Central Florida. Second, real-time intersection-approach-level crash risk was investigated by considering the effects of real-time traffic, signal timing, and weather characteristics based on 23 signalized intersections in Orange County. Third, a deep learning algorithm for real-time crash risk prediction at signalized intersections was proposed based on Long Short-Term Memory (LSTM) and Synthetic Minority Over-Sampling Technique (SMOTE). Moreover, in-depth cycle-level real-time crash risk at signalized intersections was explored based on high-resolution event-based data (i.e., Automated Traffic Signal Performance Measures (ATSPM)). All the possible real-time cycle-level factors were considered, including traffic volume, signal timing, headway and occupancy, traffic variation between upstream and downstream detectors, shockwave characteristics, and weather conditions. Above all, comprehensive real-time safety evaluation algorithms were developed for arterials, which would be key components for future real-time safety applications (e.g., real-time crash risk prediction and visualization system) in the context of pro-active traffic management.

Illegal U-turn violations are considered part of the Wrong-Way Driving (WWD) maneuvers that could result in head-on crashes and severe injuries, which are often severe because of the high speed of the approaching traffic and limited time to avoid such crash. Therefore, reviewing this type of violation and understanding the contributing factors that may lead drivers to commit such illegal maneuver would help officials foresee and consequently minimize the potential risks that could lead to WWD crashes. The purpose of this thesis is to investigate the illegal U-turn maneuvers on limited access facilities and find the significant contributing factors that encourage or discourage drivers to commit this type of violation. The study area included the Central Florida area (CF), and the South Florida (SF) area. About 6 crossover crashes and 620 citations were found at the median facilities in the study areas from year 2011 to 2016. The modeling methodology for this thesis had three goals: predicting the number of illegal U-turn violations across the traversable grass median sections per year using a Poisson regression model, selecting the most effective variables in predicting the illegal U-turn violations using the least absolute shrinkage and selection operator (LASSO) variable selection method, and estimating the probability of an illegal U-turn violation occurrence at a paved median opening for official use only per year, using a logistic regression model. To determine the variables that influence the illegal U-turn violations, 9 geometric design and 2 traffic conditions exploratory variables were analyzed in the models mentioned earlier. Several variables were found significant from the Poisson model such as the distance to the nearest interchange, the length of the median segment, the number of access points in the segment, the median design, and the speed limit. Afterwards, the LASSO method concluded that the most effective variables found were the median design and the distance of to the nearest interchange. The logistic regression model in the CF area indicated that the speed limit and the AADT as the significant contributing factors. However, in the SF area the significant variables were the distance to the nearest access point and the spacing between the median openings. The variation in results indicates a considerable difference between the two study areas that should be accounted for during the planning phases for allocating the median countermeasures. The significant variables found in the mentioned modeling approach provide a first attempt to understand the illegal U-turn violations on limited access highways, and interpret the variables which influence drivers' behavior in performing such illegal maneuver. Along with required design guidelines, the models found could be used as effective planning tools to select the appreciate locations for installing new median openings and reevaluating the existing median openings to identify locations with the lowest potential risk. Other modeling techniques that include additional factors could be tested in future research so that appropriate countermeasures can be installed to reduce or eliminate these illegal U-turns. Furthermore, the methodology could be extended to arterials (or roads with partially controlled access).