Real-Time Arousal Detection Using Galvanic Skin Response
The real-time arousal detection using galvanic skin response asset detects in real time human arousal based on measuring electro-dermal Activity (EDA), also known as skin conductance, galvanic skin response (GSR), electrodermal response (EDR), skin conductance response (SCR), and skin conductance level (SCL). EDA is related to the activity of the sweat glands, which are regulated by the sympathetic nervous system and it is measured from particular player in order to produce real-time signal features such as: mean tonic activity level, phasic activity represented by mean and maximum amplitude of skin conductance response, rate of phasic activity, SCR rise time, SCR 1/2 recovery time, and slope of tonic activity. The level of arousal may be useful for emotion detection and for adaptation purposes. The asset receives a filtered raw signal from a simple, low cost biofeedback device allowing sampling rate up to 0.8Khz. Measurements are carried out with two electrodes placed on two adjacent fingers. Recording, filtering and feature extraction might be executed on a computer (server) different than the game machine, in order to speed up all the required processing. The results are communicated from the server-side to the client component in order to be used for game adaptation.
The asset produces two main metrics featuring user arousal based on the GSR signal,
- Current level of phasic user arousal measured in N levels, i.e. from 0 to N-1, where N is user-defined and has a default value (the level is indicated by the area under the curve of the filtered GSR signal within the time window using a sampling rate chosen by the asset user). The user arousal level is determined after the first time window is expired and is based on statistical analysis of human GSR signals, thus, the accuracy is based on statistically significant data. With collecting more data from given user, the asset starts providing more accurate data.
- Current level of tonic arousal measured in N levels, i.e. from 0 to N-1, where N is user-defined and has a default value (the level is indicated by the mean amplitude of the tonic component of the signal within the time window and using a sampling rate chosen by the asset user).
As well, the asset produces real-time features of GSR signal measured from particular player such as: phasic activity represented by mean and maximum amplitude of skin conductance response (all in micro-siemens), rate of phasic activity (response peaks/sec), SCR rise time, SCR 1/2 recovery time, mean tonic activity level, and slope of tonic activity (in micro-siemens/sec).
- Sources: https://github.com/ddessy/RealTimeArousalDetectionUsingGSR
- Release: https://github.com/ddessy/RealTimeArousalDetectionUsingGSR/tree/master/LastStableVersion
- End-user documentation: https://github.com/ddessy/RealTimeArousalDetectionUsingGSR#instalation-and-usage
- Video demo: https://www.youtube.com/watch?v=fJvbghlZi1c
- Usage presentation: https://github.com/ddessy/RealTimeArousalDetectionUsingGSR/blob/master/Docs/Real-timeArousalDetectionUsingGalvanicSkinResponseAsset_Showcases.pdf
- Design: http://css-kti.tugraz.at/projects/rage/assets/designdocuments/DesignDocument-MotivationAssessmentAsset.pdf