Human trust in automation plays an important role in successful interactions between humans and machines. To design intelligent machines that can respond to changes in human trust, real-time sensing of trust level is needed. In this paper, we describe an empirical trust sensor model that maps psychophysiological measurements to human trust level. The use of psychophysiological measurements is motivated by their ability to capture a human's response in real time. An exhaustive feature set is considered, and a rigorous statistical approach is used to determine a reduced set of ten features. Multiple classification methods are considered for mapping the reduced feature set to the categorical trust level. The results show that psychophysiological measurements can be used to sense trust in real-time. Moreover, a mean accuracy of 71.57% is achieved using a combination of classifiers to model trust level in each human subject. Future work will consider the effect of human demographics on feature selection and modeling.
human-machine interface, modeling, real-time, categorical data, classifiers, discriminant analysis, human brain, intelligent machines, physiological models
Date of this Version
Hu, Wan-Lin & Akash, Kumar & Jain, Neera & Reid, Tahira. (2016). Real-Time Sensing of Trust in Human-Machine Interactions. IFAC-PapersOnLine. 49. 48-53. 10.1016/j.ifacol.2016.12.188.