Indoor Pedestrian Localization Based on Frequency Discrimination Magnetic Odometer and Dead Reckoning

Waist-mounted pedestrian dead reckoning (PDR) is widely used to provide autonomous location-based service for indoor pedestrians. However, the low-precision statistical step-size model influences the long-term positioning precision. This study proposes a frequency discrimination magnetic odometer (f-MOR)-aided dead reckoning method to address the limitation. Dual magnetometers are mounted on the front and back of the waist area. When the pedestrian walks through the indoor distorted magnetic field, the dual magnetic measurements have certain phase differences. Compared to the existing dynamic time warping (DTW)-based methods, f-MOR determines the delayed time from the more stable amplitude and phase spectra rather than directly analyzing the geometry relationship. Thus, the pedestrian front-axis velocity can be further estimated by the delayed time and the dual-sensor prior distance. To correct the step size of PDR, this study constructs the state-transition model based on the step-size scale error and f-MOR velocity error. The difference between the velocity integration and the step size is used to build the observation. This study designs a Kalman filter to realize high-precision fusion. Compared to the DTW-based methods, the experimental results show that the end-to-end positioning precision achieves 0.54%D in a 700-m indoor-outdoor environment.