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.

An intelligent transportation system (ITS) offers considerable opportunities for increasing the safety, efficiency, and predictability of traffic flow and reducing vehicle emissions. Sensors (or detectors) enable the effective gathering of arterial and controlled-access highway information in support of automatic incident detection, active transportation and demand management, traffic-adaptive signal control, and ramp and freeway metering and dispatching of emergency response providers. As traffic flow sensors are integrated with big data sources such as connected and cooperative vehicles, and cell phones and other Bluetooth-enabled devices, more accurate and timely traffic flow information can be obtained. The book examines the roles of traffic management centers that serve cities, counties, and other regions, and the collocation issues that ensue when multiple agencies share the same space. It describes sensor applications and data requirements for several ITS strategies; sensor technologies; sensor installation, initialization, and field-testing procedures; and alternate sources of traffic flow data. The book addresses concerns related to the introduction of automated and connected vehicles, and the benefits that systems engineering and national ITS architectures in the US, Europe, Japan, and elsewhere bring to ITS. Sensor and data fusion benefits to traffic management are described, while the Bayesian and Dempster–Shafer approaches to data fusion are discussed in more detail. ITS Sensors and Architectures for Traffic Management and Connected Vehicles suits the needs of personnel in transportation institutes and highway agencies, and students in undergraduate or graduate transportation engineering courses.

Crash-based safety evaluation is often hampered by randomness, lack of timeliness, and rarity of crash occurrences. Surrogate safety data are commonly used as an alternative to crash data; however, its current practice is still resource intensive and prone to human errors. The advent of connected vehicle technology allows vehicles to communicate with each other as well as infrastructure wirelessly. Through this platform, vehicle movements and signal status at the facilities can be automatically and continually monitored in real time. This study explores the viability of long-term safety performance evaluation at signalized intersection using connected vehicle technology. The development focuses on vehicle-to-infrastructure (V2I) communications which require one road-side equipment (RSE) and some level of on-board equipment to be successful. To accomplish the objective, the researchers defined useful safety measures and developed specific algorithms to derive them in real time from the V2I communication data sets. The safety measures were categorized into single-OBE measures and dual-OBE measure based on the number of the equipped vehicle needed to be monitored. We used vehicles trapped in dilemma zone as the single-OBE measure. The dual-OBE measures included rear-end and crossing conflicts. Different simulation scenarios were designed in VISSIM to test the effectiveness of the proposed framework, effect of market penetration rate as well as required observation period for effective implementation. The evaluation results indicated that the application can effectively detect changes in safety performance at full market penetration. It can detect a shift of crash pattern from rear-end crashes to right-angle crashes due to the shorted inter green interval at low traffic volume as well as the mitigation of this pattern during the medium-to-high traffic volume. The selected measures can also identify the increasing risk of rear-end and right-angle crashes after removing advance detectors at the major approaches. Sensitivity analysis from the 60 simulation hours' data showed that more than 40% and 60% penetration rate is likely to be required for a reliable detection in the low volume level and medium-to-high volume level, respectively. Increase of traffic volume activated the corresponding phases more frequently and may result in fewer safety measures being collected. Although losing the power of detection, single-OBE measure was demonstrated to be more reliable at lower penetration rate. Under low OBE market penetrations, observation period can be extended to compensate for small sample size. However, the required observation periods vary with the types of safety indicators being collected and the levels of OBE saturation. The electronic version of this dissertation is accessible from http://hdl.handle.net/1969.1/152565

In an increasingly globalised world, despite reductions in costs and time, transportation has become even more important as a facilitator of economic and human interaction; this is reflected in technical advances in transportation systems, increasing interest in how transportation interacts with society and the need to provide novel approaches to understanding its impacts. This has become particularly acute with the impact that Covid-19 has had on transportation across the world, at local, national and international levels. Encyclopedia of Transportation, Seven Volume Set - containing almost 600 articles - brings a cross-cutting and integrated approach to all aspects of transportation from a variety of interdisciplinary fields including engineering, operations research, economics, geography and sociology in order to understand the changes taking place. Emphasising the interaction between these different aspects of research, it offers new solutions to modern-day problems related to transportation. Each of its nine sections is based around familiar themes, but brings together the views of experts from different disciplinary perspectives. Each section is edited by a subject expert who has commissioned articles from a range of authors representing different disciplines, different parts of the world and different social perspectives. The nine sections are structured around the following themes: Transport Modes; Freight Transport and Logistics; Transport Safety and Security; Transport Economics; Traffic Management; Transport Modelling and Data Management; Transport Policy and Planning; Transport Psychology; Sustainability and Health Issues in Transportation. Some articles provide a technical introduction to a topic whilst others provide a bridge between topics or a more future-oriented view of new research areas or challenges. The end result is a reference work that offers researchers and practitioners new approaches, new ways of thinking and novel solutions to problems. All-encompassing and expertly authored, this outstanding reference work will be essential reading for all students and researchers interested in transportation and its global impact in what is a very uncertain world. Provides a forward looking and integrated approach to transportation Updated with future technological impacts, such as self-driving vehicles, cyber-physical systems and big data analytics Includes comprehensive coverage Presents a worldwide approach, including sets of comparative studies and applications

Traffic incidents are unforeseen events known to affect traffic flow because they reduce the capacity of an arterial corridor segment and normally generate a temporary bottleneck. Identification of retiming requirements to enhance traffic signal operations when an incident occurs depends on operations-oriented traffic signal performance measurements when effective and real-time traffic signal performance metrics are employed at traffic control centers, delays, fuel use, and air pollution may all be decreased. The majority of currently available traffic signal performance evaluations are based on high-resolution traffic signal controller event data, which gives data on an intersection-by-intersection basis but requires a substantial upfront expenditure. The necessary detecting and communication equipment also involves costly and periodic maintenance. Additionally, the full manifestation of connected vehicles (CVs) is fast approaching with efforts in place to accelerate the adaptation of CVs and their infrastructures. CV technologies have enormous potential to improve traffic mobility and safety. CVs can provide abundant traffic data that is not otherwise captured by roadway detectors or other methods of traffic data collection. Since the observation is independent of any space restrictions and not impacted by queue discharge and buildup, CV data offers more comprehensive and reliable data that can be used to estimate various traffic signal performance measures. This thesis proposes a conceptual CV simulation framework intended to ascertain the effectiveness of CV trajectory-based measures in characterizing an arterial corridor incident, such as a vehicle crash. Using a four-intersection corridor with vii different signal timing plans, a microscopic simulation model was created in Simulation of Urban Mobility (SUMO), Vehicles in Network Simulation (Veins) and Objective Modular Network Testbed in C++ (OMNeT++) platforms. Furthermore, an algorithm for CVs that defines, detects and disseminates a vehicle crash incident to other vehicles and a roadside unit (RSU) was developed. In the thesis, it is demonstrated how visual performance metrics with CV data may be used to identify an incident. This thesis proposes that traffic signals performance metrics, such as progression quality, split failure, platoon ratios, and safety surrogate measures (SSMs), may be generated using CV trajectory data. The results show that the recommended approaches with access to CV trajectory data would help both performance assessment and operation of traffic control systems. Unlike the current state of the practice (fixed detection technology), the developed conceptual framework can detect incidents that intersection-vicinity-limited does not capture detectors while requiring immediate attention.

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.

"Drivers' "illegal maneuver" and "unsafe behavior" contribute to a large number of traffic accidents every year, which are now receiving great attention from both government regulators and car manufacturers. Indeed, many research efforts have been dedicated to understanding and recognizing dangerous driving conditions to prevent crashes and injuries. In addition to the features that are already installed in the vehicles, enhancing driving safety via mobile sensing techniques (e.g., smartphones and wearables) is becoming increasingly successful with the deep penetration of smart computing. The mobile device today is equipped with numerous sensors, which has become a very effective platform to facilitate various safety applications. In this thesis, we leverage off-the-shelf mobile sensing platforms (i.e., smartphones and wrist-worn devices) to detect and analyze dangerous driving events. Our purpose is to use real-time alerts and long-term feedbacks to increase drivers' awareness of dangerous behaviors, which could help them shape good driving habits and promote safety. Specifically, two studies are presented: 1. SafeCam - analyzing intersection-related driver behaviors using smartphone sensors, and 2. SafeDrive - monitoring distracted driving behaviors using wrist-worn devices (e.g., smartwatch). The first study focuses on the intersection safety which is a critical issue in current roadway systems. In the United States, nearly one-quarter of traffic fatalities and half of all traffic injuries are attributed to intersections. We design SafeCam that uses embedded sensors (i.e., inertial sensors and cameras) on the smartphone to track vehicle dynamics while at the same time adopts computer vision algorithms to recognize traffic control information (e.g., traffic lights and stop signs). The system is able to detect dangerous driving events not only on roads but also at intersections including speeding, lane waving, unsafe turns, running stop signs and running red lights. Our second study addresses the distracted driving problem that has been considered as a major threat to the traffic safety. It is estimated that roughly 30% of vehicle fatalities involve distracted drivers, which cause thousands of injuries and deaths every year in the United States. SafeDrive is a driving safety system that leverages the wrist-worn (i.e.,smartwatch) sensors to prevent driver distractions. By tracking driver's hand motion and utilizing machine learning algorithms, SafeDrive can detect five most common distracting activities including fiddling with the control (e.g., infotainment systems), drinking/eating, using smartphones, searching items at the passenger side and reaching back seats. In the evaluation, we conduct extensive real-road experiments using different types of vehicles (e.g., sedan, minivan, and SUV) and recruiting multiple participants (15 for SafeCam and 20 for SafeDrive). The experiment results demonstrate that both SafeCam and SafeDrive are robust to real-driving environments, which could detect critical driving events and have great potential to educate drivers on how to safely operate the vehicle." --