On-body Sensing and Signal Analysis for User Experience Recognition in Human-Machine Interaction

In this paper, a new algorithm is proposed for recognition of user experience through emotion detection using physiological signals, for application in human-machine interaction. The algorithm recognizes user's emotion quality and intensity in a two dimensional emotion space continuously. The continuous recognition of the user's emotion during human-machine interaction will enable the machine to adapt its activity based on the user's emotion in a real-time manner, thus improving user experience. The emotion model underlying the proposed algorithm is one of the most recent emotion models, which models emotion's intensity and quality in a continuous two-dimensional space of valance and arousal axes. Using only two physiological signals, which are correlated to the valance and arousal axes of the emotion space, is among the contributions of this paper. Prediction of emotion through physiological signals has the advantage of elimination of social masking and making the prediction more reliable. The key advantage of the proposed algorithm over other algorithms presented to date is the use of the least number of modalities (only two physiological signals) to predict the quality and intensity of emotion continuously in time, and using the most recent widely accepted emotion model.