Motion direction estimation of walking base on EEG signal

In this paper, we propose the method to estimate the motion direction of walking based on brain wave activity. The signals of 11-channel as Electroencephalogram (EEG) of a motor area are measured to estimate the direction of a walking. After measuring the signals of 11-channel related to the motion of the lower limb, the several signal processing algorithm are applied for the signals. Specifically, we have used band pass filter to cut off the signal other than between 5-35[Hz], JADE as a form of Independent Component Analysis (ICA) and Multi-class Support Vector Machine (SVM) as a motion classifier. The interval of a motor area about the lower limb at two sides of the brain is very near and a motor area about it is located behind skin of scalp. Therefore, it is difficult for us to measure the EEG and cut off noises. However, we think that our proposal algorithm is effective to develop the real-time Brain Machine Interface (BMI). We performed the experiment that one subject lying down on a bed with closed eyes images the motion of walking, standing and turning the left and right direction among interval of 30 seconds, respectively. As the result of applying our algorithm, the recognition rate within 50-60[%] was achieved. But, it is necessary to increase number of subject, to verify the algorithm and to raise the precision of the rate. The final target is to reveal the estimation method of motion direction of walking by means of several kinds of experiments.