The Robust Heading of the Foot-Mounted PDR System Assisted by Anchor Points

Abstract

In the classic inertial-based foot-mounted PDR (Pedestrian Dead Reckoning) system, the EKF (Extended Kalman Filter) algorithm assisted by ZUPT (zero-velocity update) is widely used. However, in the actual test, the system still has a significant positioning deviation, as the ZUPT algorithm is poor in observability of system’s heading errors. In this paper, for enhancing the robustness of the PDR system’s heading, a new calibrating heading method has been proposed. It is mainly implemented by using some anchor points with known position coordinates. Firstly, since the inertial-based PDR system cannot obtain accurate initial geographic heading independently, a calibration line consisting of two known anchor points is constructed. At the initial stage, the pedestrian first walks along this line to obtain its test heading. Then using the actual heading of the line minus its test heading to get the heading offset which can be used to calibrate the initial heading effectively. Secondly, during the pedestrian movement, several known anchor points are set at the heading change points (the turning points). During pedestrian traveling, by judging whether pedestrian walk along the route between adjacent anchor points, then using the real heading of adjacent anchor points to correct the inertial recursive heading of the PDR system in real time. Also, when pedestrian passes these anchor points, by comparing the difference between the test and true coordinates of anchor points, the turning points’ position can be calibrated. Moreover, each time the pedestrian passes the anchor point, intentionally stopping for 1–2 s to guarantee in a stable stance position, then the ZARU (Zero Angular Rate Update) is used to calibrate the bias of the gyroscope. Finally, the effectiveness of heading calibration results is shown in the experiment part.