A PDR Heading Estimation Method Based on Motion Mode Recognition Using Adaptive UKF

Among indoor positioning systems, Pedestrian dead reckoning (PDR) system has been widely used for less requirement for expensive infrastructure or laborious surveys. Heading estimation is one of the important parts of PDR. There are two main sources for heading estimation. The gyroscope-based method suffers from error accumulation problem, while the method using magnetometer is vulnerable to magnetic disturbances. Therefore, a novel heading estimation method using adaptive Unscented Kalman Filter (UKF) is proposed in this paper, which fuse accelerometer, magnetometer and gyroscope on the basis of motion mode recognition. ZARU (Zero angular rate update) is utilized to estimate gyro biases and correct the gyro output based on pedestrian still/walking classification. Straight feature is then applied on the basis of straight/turning classification. Magnetometer is finally used to further reduce heading error. In addition, the adaptive adjustment mechanism of filter parameters based on the quality evaluation of measurements is designed in this paper to improve the applicability of the method to different speeds and people. Experiments have been conducted by four experimenters at three sites using two smartphones. During the experiments, the phone is waist-mounted or handheld. The results show that the 3σ positioning error of the proposed method is reduced by more than 55% compared with the gyroscope-based method and magnetometer-based method.