To classify two-dimensional motion state of step length and walking speed by applying cerebral hemoglobin information

This paper presents a research on classifying walking speed and step length simultaneously by using cerebral hemoglobin information. Nine healthy subjects performed walking task spontaneously in three levels of speed and three levels of step length. Brain information of the subjects was measured by using functional near-infrared spectroscopy (fNIRS) technology. The differences between the oxygenated hemoglobin (oxyHb) and deoxygenated hemoglobin (deoxyHb) were decomposed by wavelet packet. Feature vectors were extracted in both the time domain and frequency domain. Walking speed and step length was identified by applying support vector machine (SVM) method. The preliminary identification accuracy was 62.97%. This finding puts forward a new method for identifying two-dimensional state of lower limbs in level walking. And it lays a foundation for realizing autonomous control of walking-assistive equipment.