Driving Behavior Analysis (DBA) plays a pivotal role in designing intelligent transportation systems, enhancing road safety, and advancing Autonomous Vehicles (AVs). Driver identification, as a key aspect of DBA, has the potential to provide unprecedented opportunities for enhanced security and driver profiling. However, the current solutions for driver identification suffer from demanding extensive data collection, limited scalability, and inadequate generalization. Furthermore, DBA is also essential for training AVs, addressing the main challenges they face: accurately perceiving their surroundings to make informed decisions and to navigate safely, and effectively handling unforeseen scenarios. In the first part of this thesis, we concentrate on behavior analysis for driver identification and verification and design two novel schemes aiming to reduce data dependency and enhance the generalization ability of existing approaches. First, we propose a novel driver identification model, called DriverRep, which reduces data dependency by presenting a fully unsupervised triplet loss training. DriverRep is the first model that extracts the latent representations associated with each driver, called driver embeddings, in an unsupervised manner. In addition, we develop a novel model to tackle driver verification and impostor detection tasks based on DBA and extracted driver embeddings. In the second part, we focus on behavior prediction for AVs and their surrounding agents. First, we tackle behavior prediction in dynamic and complex scenarios by introducing three novel prediction models for forecasting drivers intentions and behaviors at unsignalized intersections. We then address social reasoning by proposing a novel prediction model that analyzes agent interactions using graph neural networks, making the scene understanding process more informative for AVs. Our proposed prediction model, called STAG, explicitly activates social modeling with a directed graph representation while considering spatial and temporal inter-agent correlations. We further design a novel prediction system, namely CAPHA, which conditions the future behavior of agents on grid-based plans modeled as a Markov decision process and solves the prediction task via inverse reinforcement learning to produce scene compliant behaviors. Moreover, we introduce a novel goal-based prediction model, called GMP, which encodes interactions between agents and dynamic and static context information to estimate the distribution of target goals, efficiently considering the inherent uncertainty in agents behavior. Extensive quantitative and qualitative comparisons have been conducted between the developed solutions and related benchmark schemes using various scenarios and environments. The obtained results demonstrate the potential of these solutions for the understudy tasks of DBA and real-world applications.

As autonomy in vehicles increases, the role of the driver will diminish, moving on to more non-driving related tasks. We are at a juncture at which cars have the ability to drive themselves, but only if the driver is ready to take over control of the vehicle when required (e.g., Tesla autopilot). Therefore, it is important that adequate alerts are used to warn drivers in various contexts to take control back from these semi-automated vehicles. Considerable research has been conducted to design the safest alerts for the takeover transition. However, more systematic research is still required to accurately predict driver responses to different parameters of the alerts. Also, takeover research has not considered drivers' states (e.g., emotions). Anger is one of the emotions that has been shown to impair driver judgment and performance. There is limited research on how anger might influence takeover performance in semi-automated driving. This study aimed to investigate the influence of anger on takeover reaction time and safety by comparing angry and neutral drivers. Additionally, the effects of increased perceived urgency of auditory alarms on takeover reaction time were measured. Data from this research was used to help test mathematical driver behavior modeling using the QN-MHP cognitive architecture. Using a motion-based simulator, 36 participants performed takeovers in semi-automated vehicle on a 3-lane highway. Between takeovers, participants performed a secondary task (i.e., online game) on a tablet. There were no significant differences in takeover reaction time between angry and neutral drivers. However, angry drivers drove faster which can lead to dangerous collisions. Angry drivers took longer to change lanes with lower steering wheel angles. Neutral drivers' slower speeds and higher steering wheel angles indicated that they initiated the lane change earlier, and thus, made safer lane changes. As expected, higher frequency and more repetitions of the auditory takeover displays led to faster takeover reaction times. QN-MHP model predictions of takeover reaction times resulted in a 68.92% correlation with the empirical data collected. The results of this study suggest that angry drivers perform riskier than neutral drivers when taking over control of a semi-automated vehicle. This study is expected to make a significant contribution to research on the influence of emotion, specifically, anger on takeover performance in semi-automated vehicles as well as takeover display design. / Master of Science / Over the last decade, there has been an increasing shift towards the automation of cars. But, this is only made possible in situations where the driver is ready to take over control of the vehicle when required (e.g., Tesla autopilot). Therefore, it is important to use the right alert sounds to warn drivers to take control back from their self-driving cars. There has been a lot of research in designing the safest alerts for taking over control of the vehicle. However, research has not considered the driver's emotions while taking over control of their vehicle. Anger has been shown to be one of the emotions that can impair driver judgment and performance. Limited research has been performed to measure how anger can influence takeover performance. This study compared how angry drivers are different from non-angry (neutral) drivers in their takeover reaction time and safety. Additionally, the effects of a more urgent sounding alert on reaction time were also measured. The data from this research help to validate the predictions of a mathematical model of driver behavior. Thirty-six participants performed takeovers in a self-driving car simulator. While they were driving in the simulator, they also played a game on a tablet. The results showed that angry drivers and neutral drivers took the same time to takeover. But, angry drivers drove faster which can lead to dangerous collisions. Angry drivers took longer to change lanes with lower steering wheel angles. Neutral drivers started changing lanes earlier because they drove slower and turned more. This meant they drove safer than angry drivers. A more urgent sounding alert led to faster takeover reaction times from both drivers. The mathematical model predictions of takeover reaction time were nearly 70% close to the actual data collected. The results of this study suggest that angry drivers perform worse takeovers than neutral drivers. The findings will help design safer alerts in self-driving cars and also contribute to the design of self-driving cars that consider the drivers' emotional states.

Highly Automated Vehicles (AVs) have the capability to revolutionize the transportation system. These systems have the possibility to make roads safer as AVs do not have limitations that human drivers do, many of which are common causes of vehicle crashes (e.g., distraction or fatigue) often defined generically as human error. The deployment of AVs is likely to be very gradual however, and there will exist situations in which the AV will be driving in close proximity with human drivers across the foreseeable future. Given the persistent crash problem in which the makority of crashes are attributed to driver error, humans will continue to create potential collision scenarios that an AV will be expected to try and avoid or mitigate if developed appropriately. The absence of unreasonable risk in an AVs ability to comprehend and react in these situations is referred to as operational safety. Unlike advanced driver assistance systems (ADAS), highly automated vehicles are required to perform the entirety of the dynamic driving task (DDT) and have a greater responsibility to achieve a high level of operational safety. To address this concern, scenario-based testing has increasingly become a popular option for evaluating AV performance. On a functional level, an AV typically consists of three basic systems: the perception system, the decision and path planning system, and vehicle motion control system. A minimum level of performance is needed in each of these functional blocks to achieve an adequate level of operational safety. The goal of this study was to investigate the effects that perception system performance (i.e., target object state errors) has on vehicle operational safety in collision scenarios similar to that created by human drivers. In the first part of this study, recent annual crash data was used to define a relevant crash population of possible scenarios involving intersections that an AV operating as an urban taxi may encounter. Common crash maneuvers and characteristics were combined to create a set of testing scenarios that represent a high iii percentage of the overall crash population. In the second part of this study, each test scenario was executed using an AV test platform during closed road testing to determine possible real-world perception system performance. This provided a measure of the error in object detection measurements compared to the ideal (i.e., where a vehicle was detected to be compared to where it actually was). In the third part of this study, a set of vehicle simulations were performed to assess the effect of perception system performance on crash outcomes. This analysis simulated hypothetical crashes between an AV and one other collision partner. First an initial worst-case impact configuration was defined and was based on injury outcomes seen in crash data. The AV was then simulated to perform a variety of evasive maneuvers based on an adaptation of a non-impaired driver model. The impact location and orientation of the collision partner was simulated as two states: one based on the object detection of an ideal perception system and the other based on the object detection of the perception system from the AV platform used during the road testing. For simulations in which the two vehicles contacted each other, a planar momentum-impulse model was used for impact modeling and injury outcomes were predicted using an omni-directional injury model taken from recent literature. Results from this study indicate that errors in perception system measurements can change the perceived occupant injury risk within a crash. Sensitivity was found to be dependent on the specific crash type as well as what evasive maneuver is taken. Sensitivities occurred mainly due to changes in the principal direction of force for the crash and the interaction within the injury risk prediction curves. In order to achieve full operational safety, it will likely be important to understand the influence that each functional system (perception, decision, and control) may have on AV performance in these crash scenarios. / Master of Science / Highly Automated Vehicles (AVs) have the capability to revolutionize the transportation system. These systems have the possibility to make roads safer as AVs do not have many of the limitations that human drivers do, many of which are common causes of vehicle crashes (e.g., distraction or fatigue). AVs will be expected to drive alongside human drivers, and so these drivers are likely to continue to be at fault in causing crashes. As part of ensuring safety, AVs will reasonably be expected to try and avoid or help reduce the severity of these crashes. AVs operate using three main systems: the perception system which consists of sensors that see the objects around the AV, the decision and path planning system, which makes decision on what the AV will do, and the vehicle motion control system. Due to the nature of the real-world, these systems may not work exactly as intended which may affect the ability of the AV to react to possible crash scenarios. Because of this, the goal of this study was to investigate the effects that perception system performance (i.e., target object state errors) has on the ability of an AV to react to crash scenarios similar to those created by human drivers. This study first defined crash scenarios using real-world crash data. A real-world perception system was then tested in these scenarios to determine object detection performance. Based on this performance, effects on safety were assessed through vehicle crash simulations. Results from this analysis showed that safety can vary based on both perception system performance and crash scenario. This highlights that it will be important to address system performance in order to achieve high levels of driving safety.

<p><a></a></p><p>Emerging technologies in transportation engineering including connected and automated vehicles (CAVs) exhibit much potential to solve a variety of persistent problems that have impaired the safety and mobility performance of transportation systems. A well-known context of such problems is the construction work zone where agencies have grappled with solutions that range from no closure, partial closure to full closure of road sections during construction, rehabilitation, or maintenance work. Road agencies also seek to develop and implement such workzone plans in a manner that does not unduly jeopardize the economic, social and environmental resources of the road users and the community where the workzone is located. In order to ensure that these three components of sustainable development are attained during road construction workzone management, road agencies seek to develop and implement tools that they can use to guide road users in a network to minimize overall delay, emissions, and fuel consumption. This thesis examines this specific context of highway administration. The thesis developed detour routing guidance for the road users in a road network with work zones in case of full closure, in a manner that is consistent with sustainable development. The research did this for the Automated vehicles (this unlikely scenario is merely considered to demonstrate the potential of connectivity in the network) and the era of connected and automated vehicles. In doing this, the thesis identified the potential benefits that CAV technology can offer in sustainable systemwide management of road work zones. The thesis considered the following sustainability-related evaluation criteria: economic (accessibility to businesses, user costs of fuel consumption, and user costs of travel delay; social (rapid access by emergency services such as ambulance); and environmental (noise pollution and Greenhouse Gas (GHG) emissions). The routing optimization was modeled as a linear programming problem and numerical experiments were carried out. The road network of Sioux Falls city was used to demonstrate the study results. The results suggest that the developed optimal sustainable routing scheme yielded significant improvement in terms of the sustainability criteria while maintaining the acceptable levels of service The results also provided insights on the prospective benefits of routing schemes developed via system optimal management (achieved through centrally-guided detour movements that is facilitated by CAV technology) vis-à-vis user equilibrium management, specifically, Nash Equilibrium.<br></p>

Sharing automated vehicles (AVs) is a possible future, where shared automated vehicles (SAVs) and pooled automated vehicles (PooledAVs) are prospective on-demand AV configurations. While SAVs and PooledAVs can contribute to the sustainability of transport systems, the success of on-demand AVs depends on whether and how the public adopts them as regular travel modes. As such, this dissertation investigates five objectives: (1) to scrutinize the essential steps of designing a future mobility survey , while the primary focus of the survey is on respondents’ intentions to adopt various AV configurations (2) to propose and validate a theoretical model for on-demand AV adoption by extending the Theory of Planned Behavior (TPB), (3) to identify the prospective use cases of SAVs as the potential precursor of on-demand AVs, (4) to identify individual characteristics that may trigger different behavioral intentions among the on-demand AV service types, and finally (5) to investigate Canadians’ intentions to adopt on-demand AVs. A nationwide Canadian survey was designed and administered in fall 2021 (n = 5002) among adults (18 to 75 years old) residing in six major Canadian metropolitan areas: Toronto, Vancouver, Ottawa-Gatineau, Montréal, Calgary, and Hamilton. The findings of this dissertation paint a complex picture of on-demand AV adoption in the Canadian context with respect to the application of constructs from common technology adoption models and will help researchers investigating the characteristics of prospective consumers of on-demand AVs to identify the importance of affective motivations regarding adopting such emerging travel modes. The results reveal that many Canadians are yet either uncertain or reluctant to adopt AV technology in shared mobility services. In this light, policymakers and planners should adjust and moderate their expectations regarding the future market for on-demand AVs and be prepared for potential changes in travel behavior by examining incremental changes in existing on-demand ride-hailing services. / Dissertation / Doctor of Philosophy (PhD) / This dissertation assesses the conditions under which Canadians are willing to use fully automated vehicles (AVs) and investigates public perceptions and intentions to use “automated ride-hailing services,” which function as a taxi or Uber/Lyft service without a driver, and “pooled automated ride-hailing services,” which are a form of ride-hailing services, where passengers share a ride with someone they do not know to save on the cost of travel. To this end, an online survey (n = 5002) was designed and administered in October and November 2021 across six major Canadian metropolitan areas: Toronto, Vancouver, Ottawa-Gatineau, Montreal, Calgary, and Hamilton. Overall, results suggest that expectations towards AVs suddenly transforming the entire transportation sector, should be moderated and “automated ride-hailing services” and “pooled automated ride-hailing services” (when they are available in the entire Canadian market) are likely to be adopted as a supplementary mobility tool rather than a substitution for current travel modes.

Les systèmes de transport intelligents (STI) sont conçus pour améliorer les transports, réduire les accidents, le temps de transport et la consommation de carburant, tout en augmentant la sécurité, le confort et l'efficacité de conduite. L'objectif final de ITS est de développer ADAS pour faciliter les tâches de conduite, jusqu'au développement du véhicule entièrement automatisé. Les systèmes actuels ne sont pas assez robustes pour atteindre un niveau entièrement automatisé à court terme. Les environnements urbains posent un défi particulier, car le dynamisme de la scène oblige les algorithmes de navigation à réagir en temps réel aux éventuels changements, tout en respectant les règles de circulation et en évitant les collisions avec les autres usagers de la route. Sur cette base, cette thèse propose une approche de la planification locale en deux étapes pour apporter une solution au problème de la navigation en milieu urbain. Premièrement, les informations statiques des contraintes de la route et du véhicule sont considérées comme générant la courbe optimale pour chaque configuration de virage réalisable, où plusieurs bases de données sont générées en tenant compte de la position différente du véhicule aux points de début et de fin des courbes, permettant ainsi une analyse réaliste. planificateur de temps pour analyser les changements de concavité en utilisant toute la largeur de la voie. Ensuite, la configuration réelle de la route est envisagée dans le processus en temps réel, où la distance disponible et la netteté des virages à venir et consécutifs sont étudiées pour fournir un style de conduite à la manière humaine optimisant deux courbes simultanément, offrant ainsi un horizon de planification étendu. Par conséquent, le processus de planification en temps réel recherche le point de jonction optimal entre les courbes. Les critères d’optimalité minimisent à la fois les pics de courbure et les changements abrupts, en recherchant la génération de chemins continus et lisses. Quartic Béziers est l'algorithme d'interpolation utilisé en raison de ses propriétés, permettant de respecter les limites de la route et les restrictions cinématiques, tout en permettant une manipulation facile des courbes. Ce planificateur fonctionne à la fois pour les environnements statiques et dynamiques. Les fonctions d'évitement d'obstacles sont présentées en fonction de la génération d'une voie virtuelle qui modifie le chemin statique pour effectuer chacune des deux manoeuvres de changement de voie sous la forme de deux courbes, convertissant le problème en un chemin statique. Ainsi, une solution rapide peut être trouvée en bénéficiant du planificateur local statique. Il utilise une discrétisation en grille de la scène pour identifier l'espace libre nécessaire à la construction de la route virtuelle, où le critère de planification dynamique consiste à réduire la pente pour les changements de voie. Des essais de simulation et des tests expérimentaux ont été réalisés pour valider l'approche dans des environnements statiques et dynamiques, adaptant la trajectoire en fonction du scénario et du véhicule, montrant la modularité du système. / Intelligent Transportation Systems (ITS) developments are conceived to improve transportation reducing accidents, transport time and fuel consumption, while increasing driving security, comfort and efficiency. The final goal of ITS is the development of ADAS for assisting in the driving tasks, up to the development of the fully automated vehicle. Despite last ADAS developments achieved a partial-automation level, current systems are not robust enough to achieve fully-automated level in short term. Urban environments pose a special challenge, since the dynamism of the scene forces the navigation algorithms to react in real-time to the eventual changes, respecting at the same time traffic regulation and avoiding collisions with other road users. On this basis, this PhD thesis proposes a two-staged local planning approach to provide a solution to the navigation problem on urban environments. First, static information of both road and vehicle constraints is considered to generate the optimal curve for each feasible turn configuration, where several databases are generated taking into account different position of the vehicle at the beginning and ending points of the curves, allowing the real-time planner to analyze concavity changes making use of the full lane width.Then, actual road layout is contemplated in the real-time process, where both the available distance and the sharpness of upcoming and consecutive turns are studied to provide a human-like driving style optimizing two curves concurrently, offering that way an extended planning horizon. Therefore, the real-time planning process searches the optimal junction point between curves. Optimality criteria minimizes both curvature peaks and abrupt changes on it, seeking the generation of continuous and smooth paths. Quartic Béziers are the interpolating-based curve algorithm used due to their properties, allowing compliance with road limits and kinematic restrictions, while allowing an easy manipulation of curves. This planner works both for static and dynamic environments. Obstacle avoidance features are presented based on the generation of a virtual lane which modifies the static path to perform each of the two lane change maneuvers as two curves, converting the problem into a static-path following. Thus, a fast solution can be found benefiting from the static local planner. It uses a grid discretization of the scene to identify the free space to build the virtual road, where the dynamic planning criteria is to reduce the slope for the lane changes. Both simulation and experimental test have been carried out to validate the approach, where vehicles performs path following on static and dynamic environments adapting the path in function of the scenario and the vehicle, testing both with low-speed cybercars and medium-speed electic platforms, showing the modularity of the system.

Today, companies face constant challenges, as the business must constantly be developed in order to compete. As new ports are being built in already existing ones, other opportunities are often made available to develop further and reduce costs and risks. Replacing some labour against automation is one of them. The study aims to ensure the risks of using selfdriving vehicles for the movement of containers in port areas. The work has been carried out at SCA in Tunadal for their new coming terminal. The aim of the study was to examine the risk associated with manpower versus self-driving vehicles exist. If, based on a safety aspect, it is profitable to go towards a more automated business. To carry out the study, documents have been collected describing the accidents in the port, two persons who are extra responsible for the accidents in the port have been interviewed and observations have been made to also see how the self-driving vehicle is to be applied to reality. When processing the data, THERP - Technique for human error rate prediction has been used for mapping the current situation. FMEA - failure mode and effects analysis and FTA - Fault tree analysis has been used to map the future situation with automation. The result shows that from a safety perspective, it is safest to use self-driving vehicles, as it is the human slips that represent the greatest risks in port areas. / Idag står företag för ständiga utmaningar då verksamheten hela tiden måste utvecklas för att kunna konkurrera. Då nya hamnar byggs i redan befintlig tas ofta andra möjligheter i förfogande för att utvecklas ytterligare och minska på kostnader och risker. Att byta ut viss arbetskraft mot automation är en av dem. Studien syftar till att se till riskerna med att använda självkörande fordon för förflyttning av containrar i hamnområden. Arbetet har genomförts hos SCA i Tunadal för deras nya kommande terminal. Målet med studien var att se till de risker som finns med att använda arbetskraft kontra självkörande fordon. Om det utifrån en säkerhetsaspekt är lönsamt att gå mot en mer automatiserad verksamhet. För att utföra studien har dokument samlats in som beskriver olyckorna i hamnen, två personer som är extra ansvariga för olyckorna i hamnen har intervjuats och observationer har gjorts för att också se till hur det självkörande fordonet ska appliceras på verkligheten. Vid bearbetning av data har THERP - Technique for human error rate prediction använts för kartläggning av nuläget. FMEA - failure mode and effects analysis och FTA - Fault tree analysis har använts för att kartlägga det framtida läget med automation. Resultatet visar att det ur ett säkerhetsperspektiv är säkrast att använda sig av självkörande fordon då det är de mänskliga snedstegen som står för de största riskerna i hamnområden.

Automated Electric Transportation (AET) is a concept of an emerging cooperative transportation system that combines recent advances in vehicle automation and electric power transfer. It is a network of vehicles that control themselves as they traverse from an origin to a destination while being electrically powered in motion – all without the use of connected wires. AET's realization may provide unparalleled returns in the form of dramatic reductions in traffic-related air pollution, our nation’s dependence on foreign oil, traffic congestion, and roadway inefficiency. More importantly, it may also significantly improve transportation safety by dramatically reducing the number of transportation-related deaths and injuries each year as it directly addresses major current issues such as human error and adverse environmental conditions related to vehicle emissions. In this thesis, a logical strategy in transitioning from today’s current transportation system to a future automated and electric transportation system is identified. However, the chief purpose of this research is to evaluate the operational parameters where AET will be feasible from a transportation operations perspective. This evaluation was accomplished by performing lane capacity analyses for the mainline, as well as focusing on the merging logic employed at freeway interchange locations. In the past, merging operations have been known to degrade traffic flow due to the interruptions that merging vehicles introduce to the system. However, by analyzing gaps in the mainline traffic flow and coordinating vehicle movements through the use of the logic described in this thesis, mainline traffic operations can remain uninterrupted while still allowing acceptable volumes of merging vehicles to enter the freeway. A "release-to-gap" merging algorithm was developed and utilized in order to maximize the automated flow of traffic at or directly downstream of a freeway merge point by maximizing ramp flows without causing delay to mainline vehicles. Through these tasks, it is the hope of this research to aid in identifying the requirements and impending impacts of the implementation of this potentially life-altering technology.

Over the past few years automotive and technology companies have made significant advances in what has been traditionally a completely human function: driving. Crash avoidance features such as lane departure warning and forward collision warning are becoming increasingly more common and cheaper to obtain, even on non-luxury vehicles. Technology companies and auto manufacturers have announced plans to have self-driving vehicles ready for public use as early as 2020. The mass adoption of automated vehicles (AVs) could significantly change surface transportation as we know it today. This thesis is intended to provide a technical analysis of the potential impacts of AVs on current light-duty vehicle miles traveled (VMT) and parking decisions, the economic desirability of widespread deployment of partially automated technologies, and methods for existing roadways to transition to connected and automated vehicle (CAV) transportation, so that policymakers can make more informed decisions during the transition to CAVs. This work takes a look at AVs from a point in time where vehicles are equipped with driver assistance systems (Level 1) to a point in time where AVs are driverless (Level 5) and can self-park. The results of this work indicate that the fleet-wide adoption of partially automated crash avoidance technologies could provide net-benefit of about $4 billion at current system effectiveness and could provide an annual net-benefit up to $202 billion if all relevant crashes could be prevented. About 25% of all crashes could be addressed by the crash avoidance technologies examined in this dissertation. Over time, as technologies become more effective and cheaper due to economies of scale, greater benefits than the $4 billion could be realized. As automated technologies become more advanced and widespread, existing roadways will need to be able to accommodate these vehicles. This work investigates the effects of a dedicated truck platoon lane on congestion on the Pennsylvania Turnpike and provides a method for existing roadways and highways to determine viable platoon demonstration sites. The initial results suggest that there are several sections of turnpike that could serve as commercial truck platoon demonstration site while still providing a high LOS to all other vehicles. Once AVs can safely and legally drive unoccupied, vehicles will no longer be limited to their driver’s destination and can search for cheaper parking in more distant parking locations. This work simulates a fleet of privately owned vehicles (POVs) in search of cheaper parking in Seattle, using a rectangular grid throughout the study area. Model results indicate that we are not likely to see significant increase in vehicle miles traveled (VMT) and energy use from cars moving from downtown parking lots to cheaper parking in distance locations but at higher penetration rates, parking lot revenues could likely decline to the point where operating a lot is unsustainable economically, if no parking demand management policies are implemented. Driverless vehicles also promise to increase mobility for those in underserved populations. This work estimates bounds on the potential increases in travel in a fully automated vehicle environment due to an increase in mobility from the non-driving and senior populations and people with travel-restrictive medical conditions. Three demand wedges were established in order to conduct a first-order bounding analysis. The combination of the results from all three demand wedges represents an upper bound of 295 billion miles or a 14% increase in annual light-duty VMT for the US population 19 and older. AV technology holds much promise in providing a more accessible and safe transportation system. This thesis can help policymakers and stakeholders maximize the benefits and minimize the challenges.

No description available.