Pedestrian dead reckoning with novel heading estimation under magnetic interference and multiple smartphone postures

Abstract Accurate heading estimation using low-cost sensors becomes an important task in pedestrian dead-reckoning (PDR). The challenges of heading estimation come from three aspects; one is the unpredictable bias originated from the built-in gyroscope sensors, which leads to the accumulative error. In addition, the magnetic disturbance and the changeable posture of the smartphone can also adversely affect the performance of heading estimation. This paper presents an improved smartphone-based PDR positioning solution (iPDR) with a novel heading estimation algorithm. The cumulative error is reduced by quasi-static constraint, and the Zero Lateral Displacement Update (ZLDU)-based optimization method. And the phone interference postures constraints are applied to eliminate the effects of multiple smartphone postures. Meanwhile, the magnetic disturbance is eliminated in the heading update stage by Kalman filter. Experimental results show that the proposed iPDR proves excellent localization performance under complex environments with severe magnetic interference and different smartphone postures, and the average heading error is within 2°, and the average positioning error is within 2 m.