Sensors system methodology for artefacts identification in Virtual Reality games

The field of affective computing has significantly emerged over the recent years, either we talk about systems that recognize and interpret human emotions or applications with assistive or therapeutic purposes that aim to treat various psychological disorders. Our goal is to design and develop a system that would aid people suffering of a common anxiety disorder of our times - the phobia. Thus, we propose a system for treating acrophobia - fear of heights - that exposes the patients in a Virtual Reality scenario depicting a mountain landscape, while their physiological signals (electrodermal activity, heart rate and respiration rate) are continuously recorded. However, during exposure, certain artefacts can appear, caused by head or hand movements, controller or mouse clicks, body rotations - and thus affect the quality of the recording. As artefact identification is a fundamental step in obtaining clean data, in this paper we present a method of recognizing artefact patterns in the signals recorded during gameplay sessions, which consists of aligning the data segments corresponding to the moments when the users performed head/hand/body movements with the artefact signals recorded during a reference procedure. The results obtained in pilot experiment with 5 users showed that the bias is lower on the perfectly aligned segments than on the segments located a few steps before and after the moment when the user introduced the artefact in the game signal, but without statistical significance. In addition, we propose a new method of estimating respiration rate based on the distance between two HTC Vive trackers placed on an elastic band, on the chest and on the upper back torso of the user.