Pedestrian Dead Reckoning Based on Activity Recognition and Stride Assessment

This paper addressed an approach of low cost pedestrian dead reckoning by using different idea from traditional dead reckoning.Activity recognition using support vector machines(SVM) was realized by analyzing the readings from the accelerometer attached to the pedestrian.And the peak detection algorithm taking advantage of alterable length sliding window was employed to estimate the real-time step frequency,Daubechies 4(DB4) wavelet base was introduced to smooth the raw data from accelerometer to enhance the performance of peak detection.Then the real-time stride was assessed as the locomotion distance according to the model of relationship between the stride and step frequency.Finally,the pedestrian dead reckoning was implemented by coupling the distance with the azimuth.And the results of experiments demonstrate the effectiveness of the proposed approach.