Estimating Walking Distance Based on Single Accelerometer

YANG Qing,CHEN Ling,CHEN Gen-cai

DOI: https://doi.org/10.3785/j.issn.1008-973x.2010.09.009

2010-01-01

Abstract:A method based on single accelerometer was proposed to estimate walking distance.With a 3-x accelerometer attached to user's crus,the method divided continuous accelerations into steps in terms of the state of crus,stance or swing,and then integrated the accelerations during swing phase twice to get the walking distance.A new self-adaptive step detection algorithm,based on the threshold step detection algorithm,could adjust walking parameters in terms of the walker's current statue (e.g.rate,stance etc).Experimental results indicated that the proposed algorithm was robust,and the effects of initial thresholds were limited.The average step error of the self-adaptive step detection algorithm was one pace.And the average walking distance errors of the self-adaptive step detection algorithm were 15.18% with constant pace and 22.34% with variable pace.While the average walking distance errors of the threshold step detection algorithm under the two situations were 31.08% and 49.82% respectively.Experimental results show that the self-adaptive step detection algorithm is more exact and robust.