Smartphone-Based Pedestrian Dead Reckoning for 3D Indoor Positioning

Indoor localization based on pedestrian dead reckoning (PDR) is drawing more and more attention of researchers in location-based services (LBS). The demand for indoor localization has grown rapidly using a smartphone. This paper proposes a 3D indoor positioning method based on the micro-electro-mechanical systems (MEMS) sensors of the smartphone. A quaternion-based robust adaptive cubature Kalman filter (RACKF) algorithm is proposed to estimate the heading of pedestrians based on magnetic, angular rate, and gravity (MARG) sensors. Then, the pedestrian behavior patterns are distinguished by detecting the changes of pitch angle, total accelerometer and barometer values of the smartphone in the duration of effective step frequency. According to the geometric information of the building stairs, the step length of pedestrians and the height difference of each step can be obtained when pedestrians go up and downstairs. Combined with the differential barometric altimetry method, the optimal height can be computed by the robust adaptive Kalman filter (RAKF) algorithm. Moreover, the heading and step length of each step are optimized by the Kalman filter to reduce positioning error. In addition, based on the indoor map vector information, this paper proposes a heading calculation strategy of the 16-wind rose map to improve the pedestrian positioning accuracy and reduce the accumulation error. Pedestrian plane coordinates can be solved based on the Pedestrian Dead-Reckoning (PDR). Finally, combining pedestrian plane coordinates and height, the three-dimensional positioning coordinates of indoor pedestrians are obtained. The proposed algorithm is verified by actual measurement examples. The experimental verification was carried out in a multi-story indoor environment. The results show that the Root Mean Squared Error (RMSE) of location errors is 1.04-1.65 m by using the proposed algorithm for three participants. Furthermore, the RMSE of height estimation errors is 0.17-0.27 m for three participants, which meets the demand of personal intelligent user terminal for location service. Moreover, the height parameter enables users to perceive the floor information.

Automated summary

This paper proposes a 3D indoor positioning method based on smartphone MEMS sensors, utilizing a robust adaptive cubature Kalman filter algorithm to estimate pedestrian heading and distinguishing pedestrian behavior patterns based on step frequency information. This method achieves accurate three-dimensional positioning coordinates of indoor pedestrians and meets the demand for location services on personal intelligent user terminals.