The classification of human lower limb motion based on acceleration sensor

In the view of different motion state corresponding to different step length in PDR of indoor and outdoor pedestrian navigation, a method is proposed to classify the lower limb motion state: walking, running, upstairs and downstairs. First use three-axis accelerometer to collect acceleration signal of waist in four conditions, then extract the inter-quartile range (IQR) and difference of peak and trough (DPT) of every step; finally build three kind of classifier include linear discriminant analysis(LDA), support vector machine(SVM) and BP neural network to classify motion states. The results show that BP neural network can be better than LDA and SVM model in classification when space between the classes is small and the average classification accuracy reached 94.5%.