Robust Determination of Smartphone Heading by Mitigation of Magnetic Anomalies

Abstract

We introduce an algorithm that provides robust three-dimensional orientation of a smartphone for pedestrian indoor localization. The algorithm focuses on integration of the magnetometer and a reformulated observation model such that the influence of magnetic anomalies is mitigated. The methodological novelty of this approach lies in the use of an extended Kalman filter (EKF), based on a state vector that contains only the slow-varying systematic deviation components of the magnetometer. We apply a statistical test to the EKF residuals to detect the presence of magnetic anomalies and update the absolute heading when beneficial conditions prevail. Otherwise, the heading is propagated based on gyroscope observations. We investigate the properties of the proposed algorithm by using simulated smartphone sensor observations with different scenarios of systematic deviations. In experiments with very accurate ground truth, the proposed algorithm achieves a root mean square error of 17.4° for the computed heading, outperforming state-of-the-art algorithms by at least 40%.