Implementation of robot–human control bio-interface when highlighting visual-evoked potentials based on multivariate synchronization index

Abstract

The introduction part, the main ways of integration of human–machine control systems integration of systems as well as the ways of visual-evoked potentials extraction in electroencephalograms (EEGs) are described. The article includes researches of electroencephalogram (EEG) signals with steady-state visual-evoked potentials for various photostimulation frequencies basing on the method of the multivariate synchronization index. Influence of length of window being processed on recognition accuracy of frequency of the signal being studied is explored. Ratio of the correctly recognized states is considered in the function of accuracy metric. Examined necessity of original signal preprocessing by way of the signal band-pass filtering. In addition, application of the multivariate synchronization index in multi-channel mode is reviewed. The result of the authors’ investigation is a number of recommendations on parameters used for extraction of steady-state visual-evoked potentials in the method of the multivariate synchronization index. Results obtained are of considerable practical importance as they can be used for brain-computer interface producing on the basis of visual-evoked potentials and later can be taken for building of control theory of robot systems of different application and for implementation of decisions on man–machine interoperability within narrow practical tasks.