Effect of Road Complexity and Explanations in shaping Situational Trust in Autonomous Driving: Implications for Autonomous Vehicle Adoption

Abstract

Decision-making of autonomous systems can be unpredictable from a human’s point-of-view, especially in complex driving situations. Existing studies identified explanation of vehicle intent as an effective method in building trust, but complexity of road interactions is a critical human concern affecting situational trust that has not been considered in prior research methods. This research studies the influence of road complexity on the efficacy of audible explanation cues to user’s situational trust of an autonomous decision-making ability and cognitive load. A mixed method within-subjects study consisting of driving video clips involving sudden vehicle slow down due to road encounters in three levels of road environment complexity (rural low traffic, highway vehicle only traffic, urban with pedestrians) and explainability (presence and absence of explanation) are tested on participants who were tasked to identify the road hazards and how the vehicle navigates through it. Participants situational trust and cognitive load are subjectively measured for each scenario while a semi-structured interview is done after the surveys to get more insights on the experience and reasoning behind their ratings. The research has found that audio “why” explanations does not have significant effect in the urban environment. The findings also suggest that higher proximity and amount of road users lowers the efficacy of explanation cues. Vehicle reaction time, speed and acceleration were also factors identified to affect user trust perceptions in all environments, while transparency of decisions was only affecting drives in rural and highway roads. In highway roads, participants expressed the need for more control while in both urban and highways, participants showed more concern in the unpredictability of road users.