Driving style characteristics and user discomfort in urban automated driving: Insights from a context-based comfort...

The perception of driving comfort resulting from differing automated driving styles has been broadly researched for both highways and rural roads, where high but steady speeds and smooth traffic flow dominate. Consequently, driving styles have been optimized for these conditions. In contrast, urban areas, where traffic and environmental complexity is much higher, challenge the design of comfortable and trustworthy driving styles. Less research has been done on urban driving situations which require almost constant adjustment of driving dynamics to cope with e.g., traffic light approaches, crossing traffic and obstacle avoidance. To address this gap, we conducted a study in a Type C fixed-base driving simulator with N = 60 participants to examine how different driving styles (dynamic vs. defensive) influence the perceived driving experience in urban situations. Psychological discomfort was continuously recorded using a handset control, complemented by standardized questionnaires (trust, comfort, acceptance, perceived safety, and behavior understanding) and post-experimental interviews on comfort aspects of the driving style. Results indicate that discomfort primarily arises from perceived uncertainty about whether and how the automated vehicle handles upcoming situations. The defensive driving style, characterized by anticipatory actions, led to lower discomfort in these situations. In contrast, the dynamic style caused increased discomfort as well as overall lower trust and acceptance ratings. Interestingly, during acceleration maneuvers, the dynamic style showed less discomfort compared to the defensive style. Our data suggest that automated vehicles should adapt driving styles contextually - favoring defensive strategies in uncertain situations (e.g., traffic light approaches) while incorporating dynamic elements where a swift response aligns with user expectations (e.g., acceleration after stops).