Increasing anthropomorphism and trust in automated driving functions by adding speech output

Abstract

Conditionally Automated Driving (CAD) functions need to be carefully examined regarding related driver attitudes such as trust and usability to increase their acceptance among future system users. By adding speech output to an existing audio-visual Human-Machine Interface (HMI), the level of trust in automation was suspected to be increased due to semantic information and the application of anthropomorphic features such as voice and gender. To test this assumption, N = 17 drivers completed two simulator drives, once with ('Speech + generic') and once without additional speech output (‘Generic’). Having interacted with the automated system in different scenarios (i.e., letting the system execute a maneuver vs. taking over control from the vehicle), drivers completed comparative questionnaires on trust, anthropomorphism and usability. The applied questionnaire on trust in automation was structured after theoretical implications and derived from previous research on trust in automation. Results showed that the ‘Speech + generic’ system was rated as superior for all three attitude measures compared to the ‘Generic’ system. Furthermore, the present study describes a first approach on comprehensively examining trust in vehicle automation. We brought forth evidence that speech output is highly relevant in order to improve driver attitudes that affect acceptance of automated systems.