Creating Appropriate Trust for Autonomous Vehicles – a framework from HMI design

While autonomous vehicle technology progresses, potentially leading to a safer traffic environment, many challenges remain within the area of human factors. One very important factor that must be addressed is to what extent the driver (user) will be able to trust the self-driving car and its technology. Trust is a cornerstone in the public acceptability of this new and innovative technology. The aim of the thesis is to explore how an appropriate level of user trust for future autonomous driving vehicles is created. In order to do so, a greater understanding of what affects trust in the Human-Machine Interaction (HMI) in autonomous driving vehicles is studied. This knowledge is then used to generate a guiding framework for implementing trust-related factors into the interaction system. This framework is then used to create an example concept of how a human-machine interaction could be created with regards to trust. The first main result of the project work is a trust-based framework based on a driving scenario, which can aid future human machine interaction designers in creating interaction systems focused on generating appropriate driver trust. As an initial attempt to corroborate the framework, one feature – object recognition – is tested and validated through a validation test of several test concepts, which confirmed the framework’s usefulness in guiding a trust-based development process. The second main result, based on the validation test results, is the further development of an illustrated example concept; demonstrating what types of trust-based interaction system concepts the framework can be used to create. The authors recommend that HMI designers and autonomous vehicle manufacturers take a more holistic perspective on trust rather than focusing on single, “isolated” events; for example, understanding that trust formation is a dynamic process that starts long before a user’s first contact with the system, and continues long thereafter. Furthermore, factors affecting trust change, both during an interaction and over time; thus HMI concepts need to be able to adapt. Future work should be dedicated to understanding how trust-related factors interact with each other, as well as on more comprehensively validating and testing the trust-based framework developed in this thesis.

Fredrick Ekman. Mikael Johansson
Research area
Systems for accident prevention and AD
Publication type
Master's thesis
Year of publication