On preference modeling techniques to analyze autonomous vehicle adoption and use

In the past decade, there have been rapid advancements in autonomous driving technologies that enable vehicles to drive by themselves. Vehicles equipped with such technologies are commonly referred to as autonomous vehicles (AVs). These technologies have already reached a stage of maturation, and are being pilot-tested as well as tested on public roads. Despite these advancements, much is still unclear regarding preferences for this technology and potential AV use implications. In this dissertation, advanced econometric models are used for gaining insights regarding individual propensities to accept and use AVs, as well as the implications of such use on the transportation infrastructure. Before AVs can gain widespread acceptance, it is critical to demonstrate (to regulatory authorities and the general public) the safe and reliable operation of AVs. Such a demonstration would involve extensive testing of AVs on public roads. For the success of such tests, it is important to ensure that people do not perceive the operation of these vehicles on public spaces as unsafe. This dissertation will make use of a nationwide survey conducted by the Pew Research Center to understand the socio-demographic characteristics and other attributes that affect an individual’s perception of the safety of sharing the road with AVs. A simultaneous equation model is estimated that takes into account the endogeneity between factors affecting individuals’ acceptance of AVs and their perceived safety of sharing the road with AVs. Once AVs have been deemed safe and acceptable for public use, services based on AVs may be made available to consumers in a variety of formats. Some households may choose not to own a vehicle and instead access AV-based services through a mobility-as-a-service (MaaS) framework. To understand ownership preferences regarding AVs, a recent survey conducted in the Puget Sound Region that elicited preferences for different paradigms of AV use and ownership is used. The stated preferences are translated to a ranking format, and the underlying assumptions and robustness of different ranking models based on the utility maximization principle are investigated. Model results on the preferred method of AV adoption suggest that a considerable proportion of individuals would opt to use AVs in the context of ride-hailing and by extension would be willing to accept a MaaS framework that utilizes AVs. Ride-hailing is a service that has been rapidly gaining popularity over the past decade. The preference for AV-based ride-hailing is expected to accelerate this trend. Unfortunately, most of the current travel demand modeling frameworks (TDMs) are not adequately structured to handle ride-hailing trips. A specific aspect of ride-hailing trips that the TDMs currently fail to capture is deadheading trips. In this dissertation, a framework and an approach to model deadheading trips are proposed. Data obtained from a ride-hailing service provider is employed for model estimation. Overall, the research undertaken as part of this dissertation will provide transportation planners and safety professionals with insights and guidance on how best to navigate the transition into the era of autonomous vehicle adoption and use.Civil, Architectural, and Environmental Engineerin