Estimation of upper-limb motion in sagittal plane based on EEG signals

Abstract

To control a wearable robot, a surface electromyogram (sEMG) signal is one of the widely used biological signals. However, there are some persons who cannot measure necessary sEMG signal. An electroencephalogram (EEG) signal is expected to serve as one of the alternatives to a sEMG signal. An EEG signal has the advantage that the signals can be measured from more persons in comparison with a sEMG signal. However, the measured EEG signals may be not always directly related with the user's motion intention. The purpose of this study is to control the wearable robot according to the user's motion intention by using EEG signals. In this paper, we try to estimate user's upper-limb motion in the sagittal plane. In the estimation, relatively simple signal processing is used. Since there is a lot of individual difference in an EEG signal, a certain learning method is necessary to adjust the individual difference in the estimation method. In the learning, two kinds of learning processes are performed. In the first case, the subject moves own upper-limb during the learning, and compares the motion of the upper-limb and the estimation results. On the other hand, in the second case, the subject does not move the upper-limb and only imagines the upper-limb motion. Each estimation result is evaluated by performing the experiments.