A Trust Model for Human-Machine Systems under Mutual Influence of Human and Machine Reliability: Virtual Reality Use-cases

Human trust in machines plays a pivotal role in performing a task effectively within human–machine systems (HMS). Prior work defined a three-layered framework for such trust, encasing all three elements of HMS – human, machine, and environment, but also indicated two deficiencies. Firstly, there is an absence of a model with metrics spanning all layers to objectively measure fluctuations of the trust (trust dynamics) in real time. Secondly, there is an inadequate consideration of human and machine reliability for evaluating task performance and the trust. Herein, the aim of this thesis is to propose and evaluate a trust model with metrics spanning all three layers to objectively measure trust dynamics. To achieve this aim, two challenges must be addressed: (1) assessing the mutual influence of human and machine reliability on task performance and (2) formulating the metrics based on objective measurement of behavioral data. By addressing these challenges, this thesis makes three contributions: (a) developing a trust model with metrics aligned with the three-layered framework and formulated to be objectively measurable; (b) enabling concurrent consideration of human and machine reliability within HMS; and (c) investigating the mutual influence of human and machine reliability on task performance and applying these findings in the model’s empirical evaluation across two virtual reality (VR) use-cases. The outcomes of the evaluation demonstrate the pertinence of the model in measuring trust dynamics while accounting for both human and machine reliability. And the model’s objective measurement was notably sensitive to trust dynamics than its subjective counterpart. The proposed model could enable designing trustworthy and adaptive HMS for applications across various domains such as autonomous vehicles, telesurgery, aviation, and immersive VR training.