User comfort and naturalness of automated driving: The effect of vehicle kinematics and proxemics on subjective response

Abstract

Higher-level Automated Vehicles (AVs, SAE Level 4+) need to provide a comfortable user experience to enhance public acceptance. AV driving styles, characterised by vehicle kinematics and proxemics, affect user comfort, with “human-like” driving styles expected to provide natural feelings to further improve user comfort. This study investigated how the kinematic and proxemic characteristics of an AV’s driving style affect user comfort and naturalness of a ride. The similarities in automated and users’ own manual driving style, and how these similarities affect evaluations, was also investigated. Using a motion-based driving simulator, participants experienced three Level 4 automated driving styles (Defensive, Aggressive, and Machine-Learning based), and a manual drive. Participants provided ratings (separately) for comfort and naturalness of each automated controller, as it negotiated twenty-four UK road sections, with varying geometric and roadside features. Linear mixed-effects models were used to examine the effect of kinematics and proxemics of the AV’s driving style, on subjective evaluation of comfort and naturalness of the ride, and how similarities between users’ own driving style and that of the AV affected riders’ evaluation. Results showed that the AV controllers’ lateral and rotational kinematics significantly influenced both comfort and naturalness, while longitudinal jerk only affected comfort. The Euclidean distance in a range of kinematics, characterising similarities between manual and automated driving styles, had varied effects on subjective evaluations. This research facilitates understanding how control features of AVs affect user experience, which will contribute to designing more user-centred controllers, leading to better acceptance of higher-level AVs.