Modelling and control of human response to a dynamic virtual 3D face

Abstract

This chapter of the monograph introduces the application of identification and predictor-based control techniques for modelling and design control schemes of human response as reaction to a dynamic virtual 3D face. Two experiment plans were used, the first one – 3D face was observed without virtual reality headset and the second one – with virtual reality headset. A human response to the stimulus is observed using EEG-based emotion signals (excitement and frustration). Experimental data is obtained when stimulus and response are observed in real time. Cross-correlation analysis of data is demonstrated that exists dynamic linear dependency between stimuli and response signals. The technique of dynamic systems identification which ensures stability and possible higher gain for building a one-step prediction models is applied. Predictor-based control schemes with a minimum variance or a generalized minimum variance controllers and constrained control signal magnitude and change speed are developed. The numerical technique of selection the weight coefficient in the generalized minimum variance control criterion is based on closed-loop stability condition and admissible value of systematic control error. High prediction accuracy and control quality are demonstrated by modelling results