Brain-robot interface: Distinguishing left and right hand EEG signals through SVM

Abstract

In this paper a new method of implementing brain-robot interface is presented. Motor imagery (MI) is kind of spontaneous EEG that is employed into the EEG-based BMIs. The features extraction and classification of EEG data related to the left and right hand motor imagery are performed. At first, the EEG signals from six channels are collected, and then filtered by low-pass filter. Wavelet transform decomposes the signal into frequency sub-bands as features. In the next step, support vector machine (SVM) classifies features in two classes: left or right hand motor imagery. The classification accuracy rate is 75%. Finally the output of classification is applied to move the arm of Tabriz-Puma robot.