Towards a Personalized Trust Model for Highly Automated Driving

Abstract

User acceptance of automated vehicles (and dependent dimensions such as road safety, frequency of use or level of recommendation) is said to be highly dependent on the operator’s individual trust in this technology. As a consequence, the development of driving functions and future driver-vehicle interfaces should allow for appropriate trust calibration. To better understand trust and the effect of mis-calibration on the way to a personalized trust model, we propose a set of trust-related research questions derived from related work and our own user studies. Based on preliminary investigation, we recommend examining 1) differences in users and subgroups of users, 2) different levels of trust based on situation or context, 3) methods for quantifying trust in naturalistic driving studies, and 4) definitions for an established/approved trust model and the individual calibration of the model with regard to driving behavior and automotive user interfaces. The final outcome should be a multidimensional trust model that fits the individual passenger/driver by dynamically adapting driving mode and UI representation/feedback.