Boyle, LindaMiller, Erika Elizabeth2018-07-312018-07-312018-07-312018Miller_washington_0250E_18694.pdfhttp://hdl.handle.net/1773/42254Thesis (Ph.D.)--University of Washington, 2018Autonomous vehicle systems have elicited the attention of car manufacturers, consumers, policy makers, and the media as they offer societal, environmental, and economic benefits. However, prolonged exposure to these systems may lead drivers to adapt to these systems in ways not anticipated by the designer; resulting in unintended safety consequences. To explore this issue, a longitudinal driving simulator study was conducted to evaluate behavioral adaptations due to exposure to an active lane keeping system. In this study, performance before, during, and after exposure to the semi-autonomous system was compared. Forty-eight participants (30 treatment, 18 control) completed a series of eight drives across three separate days. Treatment participants were exposed to approximately 40 minutes of baseline [manual] driving, 80 minutes of semi-automated driving, and 40 minutes of post-automated [manual] driving. A control group was exposed to approximately 160 minutes of manual driving, but otherwise identical study procedures, in order to provide a reference for time on task effects. Changes in secondary task engagement (number completed and accuracy), driving performance (SDLP and TTC), cognitive workload (TDRT response time and miss rate) and eye glance behavior (mean glance duration, 90th percentile glance duration, total eyes-off-road time, and percent long glances) were modeled using generalized linear mixed models. Cluster analysis techniques were used to examine the effects of trust in automation on behavioral adaptations. The findings of this dissertation suggest that drivers began to rely on automation for support and experienced adverse effects when the system was removed. Moreover, drivers with higher self-reported trust in the autonomous system experienced the largest degradations in performance and were associated with inherently more risky driving habits. By identifying the associations between trust and behavioral adaptations over time, vehicle systems, infrastructure, and educational programs can be designed to support appropriate use and attention allocation, in order to minimize adverse effects during handover and takeover of vehicle control.application/pdfen-USnoneautomationbehavioral adaptationscognitive workloaddriver behaviordriving performancetrustTransportationBehavioral psychologyCivil engineeringCivil engineeringThesis