A pedestrian dead reckoning system using a foot kinematic constraint and shoe modeling for various motions

Abstract In this paper, we propose a pedestrian dead reckoning (PDR) system that uses the velocity predicted by a foot kinematic constraint and ellipsoidal shoe modeling in the contact phase. Generally, a PDR system with foot-mounted inertial sensors often uses zero-velocity update (ZUPT) to reduce the influence of the bias and white noise in the gyroscope and accelerometer signals. However, when a pedestrian walks irregularly, ZUPT is unsuitable because the velocity is not small enough to assume that it is zero during the stance phase. Unlike a conventional PDR system that uses ZUPT in the stance phase when the shoe has completely stopped, the proposed PDR system uses the predicted velocity in the contact phase when the shoe touches the ground from the heel-strike to the toe-off states by employing the lever arm vector and the angular velocity to predict the nonzero velocity under the kinematic constraint. The lever arm vector is obtained by the attitude of the shoe according to ellipsoidal shoe modeling. The predicted velocity is used as a measure of the filter during the contact phase. The proposed PDR system can be utilized in various motions with nonzero velocity in the contact phase. The performance of the algorithm was verified by experimental results with various motions.