Pedestrian tracking through inertial measurements : a data-fitting approach to minimise velocity drift

Abstract

Most modern outdoor tracking systems utilize GPS technology. However, typical pedestrian environments tend to distort and block satellite signals, thereby making GPS communication systems unreliable. This is particularly true for indoor environments. An alternative pedestrian dead reckoning system based solely on inertial sensors is presented. Noisy acceleration signals result in a heavy velocity drift. This study proposes staticperiod and dynamic-period compensation methods that would allow for accurate 2-D positioning to be estimated. Data fitting techniques are used to reset velocity drift error after every step, thereby preventing accumulation of errors, as well as to compensate for the errors incurred during the step itself. Results show that the combined system is able to track rectilinear motion under normal walking conditions with an average accuracy of 1.4%, while various lengths of non-linear 2-D paths travelled by different users with varying gaits, paces and directions were successfully tracked with a maximum error of 5%.