The ubiquity of smartphones has made a remarkable influence on everyone's day to day life. Variety of useful built-in sensors provide smartphones with a convenient floor for data collection and analysis. Application development based on the user's location and movement is not a difficult task nowadays. But injuries and deaths due to smartphone-distracted movement on roadways is on the increase. This study explores the capabilities of smartphone inertial sensors for pedestrian activity recognition. Smartphone distracted movements can be predicted from the associated pedestrian's posture, thus inertial sensors can provide effective solution for this specific task. Volunteers were asked to perform different pedestrian activities with smartphones in their hand or in trouser pocket. Accelerometer and gyroscope data were collected, and time windowing was applied for proper segmentation of the data. After time and frequency domain feature extraction of these segmented data streams, two classical supervised machine learning approaches (SVM and Random Forest) were undertaken for correct prediction of seven different pedestrian activity labels. Furthermore, we implemented a deep learning classifier (CNN) for direct activity recognition using raw data. The training and testing procedure includes three types of systems: single-subject, all-subject, and leave-one-subject-out models. For performance evaluation, we used the F-score metric, which can reach up to 92.3%, 98.1% and 97.2% for these three models, respectively. CNN with raw data provides much better accuracy than the classical machine learning models. With the capability to identify pedestrian activity and thus distracted pedestrians with great accuracy, our approach lays the foundation for a smartphone application based real time P2V warning system. In this system, the vehicle's driver gets a warning in his smartphone about the nearby presence of a distracted pedestrian.

Complete street systems integrate a wide range of users in the same space, with unequal risks and responsibilities. This makes driver attention a critical factor in assuring the safety of vulnerable users. The Conditioned Anticipation of People psychological model of driver attention proposes that drivers reflexively reengage their metacognitive processes when they anticipate visually interacting with the human face or form due to the neurological priority that the brain places on human recognition. To test this model, an eye-tracking tabulation was generated from the SHRP2 Naturalistic Driving Study that measured midsegment percent of time on-task and multitasking behavior for 200 sites in Tampa, Florida and Seattle, Washington. This attention data was statistically analyzed for the impacts of a wide range of context variables using single variable ANOVA and various multivariate models such as ordered probit fractional split and ordered probit models. Context features with a strong correlation to vulnerable user presence that support driver's visual recognition of that presence were also strongly correlated with driver attention. Features like corridor width, block length, doorway density, and sense of enclosure had the largest impact. Features that did not have an impact on the potential visual connection with street users, like lane width, right of way width, onstreet parking, functional classification, or Walkscore had no impact on driver attention or weak effect sizes, despite strong correlations with vulnerable user presence. Crash history was evaluated in conjunction with the variables most sensitive to driver attention with mixed results. Many of the features that increase the potential for drivers to see and interact with people also contribute to increases in vehicle to vehicle conflicts. A decrease in crash rate with increasing sidewalk width implies that the CAP effect can have some impact on crashes. Implications for complete streets and community design are discussed.

Several unconventional intersection designs were proposed and implemented to enhance traffic safety and operation at intersections. The efficiency of these intersection designs was not sufficiently evaluated in the previous research because of the limited implementation of such designs. However, with the growing interest in the implementation of unconventional intersections by municipalities and transport agencies, it has become a need for a comprehensive evaluation of their safety and operational benefits. Therefore, this dissertation aims to evaluate the safety and operational aspects of unconventional intersection designs by employing different research approaches: crash analysis, microscopic simulation, and driving simulation. Firstly, this dissertation evaluated the effectiveness of median U-turn crossover-based intersections (median U-turn (MUT) and restricted crossing U-turn (RCUT) intersections), which have the least number of traffic conflicts among other unconventional intersection designs, in enhancing traffic safety by estimating crash modification factors (CMF) for their implementation. The results indicated that MUT and RCUT intersections are safer than the 4-leg conventional intersection. Secondly, A new innovative intersection design, which has been given the name "Shifting Movements" (SM) intersection, was introduced and proposed to replace the implementation of the RCUT intersection under moderate and heavy minor road traffic conditions. Evaluation of the operational benefits of this intersection design was performed in the microscopic simulation environment by assuming different traffic volume levels and left-turn proportions to represent the peak hour with moderate to high left-turn traffic volumes. The results demonstrated that the SM intersection design significantly outperforms conventional and RCUT intersections when they are subjected to high traffic volumes in terms of average control delay and throughput. Finally, A driving simulation experiment was conducted to evaluate the safety aspects of the SM intersection design. Several surrogate safety measures were adopted for the evaluation. The effectiveness of using infrastructure-to-vehicle (I2V) communication for mitigating the confusion at unconventional intersections has been also evaluated in this research. Findings indicated that RCUT and SM intersections have similar safety performance and crossing them is safer than crossing the 4-leg conventional intersection. It was found that using I2V communication is helpful in understanding unconventional movement patterns. This dissertation can be a solid reference for decision-makers regarding the implementation of unconventional intersection designs.

Urban arterials connect multiple areas in the city and encourage non-motorist activities. Hence, the safety and operations on urban arterials is vital as they improve the mobility of daily commuters and road users. This research aims to facilitate traffic operations on urban arterials by proposing multiple mythological approaches to estimate and predict turning movement counts at signalized intersections using traffic data from adjacent intersections. Further, it aims to improve the safety by developing crash prediction models, identifying the hotspots for multiple crash types, and indicating the factors contributing to operating speed as well as non-motorist crashes. The analyses included tuning, testing, and comparing multiple parametric and machine learning models. First, a framework was proposed to estimate cycle-level turning movements' counts at signalized intersections based on traffic data from adjacent intersections. As a result, generic Extreme Gradient Boosting (XGBoost) models were developed to estimate through and left turn movements with Mean Absolute Error (MAPE) 9.53% and 4.7%, respectively. Afterwards, multiple machine learning models were trained and compared to predict through and left turning movements. The GRU models outperformed other developed models and were able to provide accurate time horizon predictions for five cycles in the future. The developed models for estimation and prediction could emulate detection systems at signalized intersections, improve traffic signals optimization, assist in corridor management and safety, and save capital by eliminating the need for equipment investment at many intersections. On the other hand, aiming to improve road users' safety on urban arterials, this research proposed an integrated approach to identify the hotspots by developing crash prediction models (Safety Performance Functions) considering context classification. The study utilized big data and compared a wide array of statistical and machine learning models that were developed to estimate reliable non-motorist exposure. The results indicated that XGBoost is the best model to estimate non-motorist exposure at intersections and along the roadway segments. Further, the proposed approach included developing Safety Performance Functions (SPFs) to identify the hotspots for two types of crashes (i.e., vulnerable road users' crashes at intersections and bike crashes along the road). It was found that the hotspots were more likely to be located near the city of Orlando. Coastal roadways were classified as cold categories regarding bike crashes. Finally, the research investigated the factors contributing to operating speed considering context classifications. The analyses indicated the significant factors that influence operating speed or non-motorist crashes such as average block length, shoulder width, proportion of population below poverty, and number of signalized intersections per mile. It also illustrated the potential speed management countermeasures that significantly influence the operating speed. These countermeasures could have potential influence on roadway safety if implemented. This could help the decision makers to determine the best countermeasures to be implemented along roadway segments for different context classification roads.

The recent development in communication technologies facilitates the deployment of connected and automated vehicles (CAV) which are expected to change the future transportation system. CAV technologies enable vehicles to communicate with other vehicles through vehicle-to-vehicle (V2V) communications and the infrastructure through Vehicle-to-infrastructure (V2I) communications. Since the real-world CAV data is not currently available as of today, simulation is the most commonly used platform to evaluate the future V2X system. Although several studies evaluated the effectiveness of CAVs in a small roadway network, there is a lack of studies analyzing the impact of CAVs at the network level by considering both freeways and arterials. Also, none of the previous studies have attempted to differentiate the benefits of CAVs over only automated vehicles (AVs) by incorporating multiple preceding vehicles' information (i.e., acceleration, position, etc.). On the other hand, most of the simulation-based studies assumed the uninterrupted communication between vehicles in the CAV environment which might not be feasible in reality. Hence, there is still a research gap that exists for which this study tried to fill this gap. Therefore, this study developed a calibrated and validated large-scale network for the deployment of CAV technologies by utilizing Dynamic Traffic Assignment (DTA) model in Orlando metropolitan area, Florida, using Multi-Resolution Modeling (MRM) technique. Also, the study proposed a signal control algorithm through V2I technology in order to elevate the performance of CAVs at intersections. Different car-following models were utilized to approximate different CAV technologies (CAV, AV, and CV (connected vehicle)) in the simulation environment. Hence, the study analyzed the benefits of CAV over AV with different market penetration rates (MPRs). Furthermore, the study considered the performance of different communication system along with the traffic condition by utilizing Dedicated Short-Range Communications (DSRC or IEEE 802.11p) and wireless access (IEEE 1609 protocol) for the application of vehicle ad-hoc network (VANET). To this end, the study evaluated the safety effectiveness of different communication protocols under the CAV environment. Aimsun Next and SUMO & OMNET++ based Veins simulator were used as the simulation platform. Different car-following models, signal control algorithm, and communication systems were coded by using the application programming interface (API) and C++ language. For the traffic efficiency, the study utilized travel time and travel time rate (TTR) while for the safety evaluation, different surrogate safety measures; speed, and crash-risk models were used. Also, several statistical tests (e.g., t-test, ANOVA) and modeling techniques (e.g., generalized estimating equation, logistic regression, etc.) were developed to analyze both safety and mobility. The results of this study implied that CAV could improve both safety and efficiency at the network level with different MPRs. Also, CAV is more efficient compared to the only AV in terms of both traffic safety and mobility. Different communication protocols have a significant effect on traffic safety under the CAV environment. Finally, the results of this study provide insight to transportation planners and the decision makers about the benefits of CAV at the network level, different CAV technologies, and the performance of different communication systems under the CAV environment.

In recent years, cycling has become an increasingly popular transportation mode around the world. In contrast to other popular modes of transportation, cycling is more economic and energy efficient. While many studies have been conducted for the bicycle safety analysis, most of them were limited in terms of bicycle exposure data and on-street data. This study tries to improve the current safety performance functions for bicycle crashes at urban corridors by utilizing crowdsource data from STRAVA and on-street speed management strategies data. Speed management strategies are any roadway alterations that causes a change in motorists' driving behavior. In Florida, these speed management strategies were defined by the Florida Department of Transportation (FDOT) Design Manual. Considering the disproportion in the representation of cyclists from the STRAVA data, adjustments were done to more accurately represent the cyclists based on the video detection data by developing a Tobit model. The adjusted STRAVA data was used bicyclist exposure to analyze bicycle crashes on urban arterials. A Bayesian joint model was developed to identify the relations between the bicycle crash frequency and factors relating speed management strategies. Other factors such as vehicle traffic data, roadway information, socio-demographic characteristics, and land use data were also considered in the model. The results suggested that the adjusted STRAVA data could be used as the exposure for bicycle crash analysis. Also, the results highlighted the significant effects of speed management strategies such as parking lots and surface pavement. It is expected that the results could help engineers develop effective strategies to enhance safety for bicyclists.

A multitude of externalities affects transport efficiency and numbers of trips. Population expansion, urban development, political issues, fiscal trends, and growth in the field of connected, automated, shared, and electric (CASE) vehicles have all played prominent roles. While the market is keenly aware of the upcoming shift to the CASE vehicles, the transformation itself is reliant upon the development of technologies, customer outlook, and guidelines. The purpose of this research is to establish an overview of the possible network design problems, as well as potential consequences to vehicle automation systems by employing machine learning and system dynamics analysis. Finally, the cost of the required highway expansion for the critical links in the traffic network will be predicted. First, model was created for calculating traffic flow activity and necessitated highways to consider the impact of CASE vehicles between 2021 and 2050. Second, an economic evaluation outline was created to calculate optimum time and roadway improvement scenarios by a cost-prediction model using machine learning. Florida's interstate highways were employed as the subjects for the case study. The research showed that non-linear models had a better ability in the estimation of traffic flow, while linear models were better predictors of highway construction cost. These results also showed new technologies would add to traffic flow and capacity, with the increase in flow outpacing the increase in capacity. The consequences of this would be the level of service (LOS) of the current infrastructure decreasing. This study's results can assist discussion at the national and local level between government, networkers, automotive companies, tech-providers, logistics companies, and stakeholders for whom the practicality provided by the transportation infrastructure is crucial. This allows executives to create effective guidelines for subsequent transportation networks, ultimately accelerating the CASE vehicle network rollout to increase our current road network's level of service.

Nowadays, technology is employed in many safety applications and countermeasures that would enhance traffic safety by influencing some crash-related factors. Therefore, crash-related factors must be determined for every roadway element by the development of safety performance functions. Safety performance functions (SPF) are employed to predict crash counts at the different roadway elements. Several SPFs have been developed for the various roadway elements based on different classifications such as functional classification and area type. Since a more detailed classification of roadway elements leads to more accurate crash predictions, multiple states have developed new system to categorize roads based on a comprehensive classification. In Florida, the new roadway context classification system incorporates geographic, demographic, and road characteristics information. In this study, SPFs have been developed in the framework of the FDOT roadway context classification system at three levels of modeling, context classification (CC-SPFs), area type (AT-SPFs), and statewide (SW-SPF) levels. Crash and traffic data of 2015-2019 years have been obtained. Road characteristics and road environment information have been also gathered along Florida roads for the SPF development. The developed SPFs showed that there are several variables that influence the frequency of crashes, such as annual average daily traffic (AADT), signalized intersections and access points densities, speed limit, and shoulder width. However, there are other variables that did not have an influence on crash occurrence such as concrete surface and the presence of bicycle slots. CC-SPFs had the best performance among others. Moreover, network screening to determine the most problematic road segments has been accomplished. The results of the network screening indicated that the most problematic roads in Florida are suburban commercial and urban general roads.

Understanding the e-bicycle overtaking behaviors is very important for analyzing the bicycle traffic and designing the automated driving system in the future. However, currently, few solid insights of overtaking behaviors were made, due to a lack of efficient models to simulate realistic bicycle overtaking trajectories, and especially, the tactical strategies. This paper referred to the latest advancement in the automated driving system, and applied a maximum entropy based inverse reinforcement learning method to repeat realistic bicycle overtaking trajectories, and identify tactical strategy preferences and considerations of trade-off factors (safety, comfort, and efficiency) during an overtaking task. The method was demonstrated on thirty-five e-bicycle overtaking events, and successfully generated overtaking trajectories of similar features (acceleration, jerk, speed, lane deviation, and collision avoidance) as observed trajectories. The results show that, in general, during an overtaking task, the feature that is first considered is the safety-related factor, the second is the speed control and the third is the lateral movement; nevertheless, there are significant individual heterogeneities when deciding overtaking behaviors; in addition, the overtaking behavior is also affected by the type of overtaken bicycle. Implications for research and practice are proposed in this study.

Recently, with the development of connected vehicles and mobile sensing technologies, vehicle-based data become much easier to obtain. However, only few studies have investigated the application of this kind of novel data to real-time traffic safety evaluation. This dissertation aims to conduct a series of real-time traffic safety studies by integrating all kinds of available vehicle-based data sources. First, this dissertation developed a deep learning model for identifying vehicle maneuvers using data from smartphone sensors (i.e., accelerometer and gyroscope). The proposed model was robust and suitable for real-time application as it required less processing of smartphone sensor data compared with the existing studies. Besides, a semi-supervised learning algorithm was proposed to make use of the massive unlabeled sensor data. The proposed algorithm could alleviate the cost of data preparation and improve model transferability. Second, trajectory data from 300 buses were used to develop a real-time crash likelihood prediction model for urban arterials. Results from extensive experiments illustrated the feasibility of using novel vehicle trajectory data to predict real-time crash likelihood. Moreover, to improve the model's performance, data fusion techniques were proposed to integrated trajectory data from various vehicle types. The proposed data fusion techniques significantly improved the accuracy of crash likelihood prediction in terms of sensitivity and false alarm rate. Third, to improve pedestrian and bicycle safety, different vehicle-based surrogate safety measures, such as hard acceleration, hard deceleration, and long stop, were proposed for evaluating pedestrian and bicycle safety using vehicle trajectory data. In summary, the results from this dissertation can be further applied to real-time safety applications (e.g., real-time crash likelihood prediction and visualization system) in the context of proactive traffic management.