A particle filter for 2D indoor localization relying on magnetic disturbances and magnetic-inertial measurement units

Abstract

Distortions of the Earth's magnetic field are critical when a magnetic sensor is used for aiding the orientation estimation, especially indoors. However, based on suitable maps of the magnetic field, the spatial patterns of indoor magnetic fields may aid the position estimation. In this paper, we propose a method based on a Particle Filter (PF) for two-dimensional (2D) localization of a human walker wearing a Magnetic/Inertial Measurement Unit (MIMU) at the waist. Magnetic disturbances in the global reference frame and pelvis rotation are estimated using a Kalman-based orientation estimator. The pelvis rotation and the output from an accelerometer-based step detector drive the propagation of the particles; the estimated magnetic disturbances are looked up in the magnetic map for weight update. The proposed PF-based 2D localization method was tested using simulated and real MIMU data collected from a healthy subject walking for three minutes along eight-shaped paths within a 3×2 m area. Using reference data from an optical motion capture system, the 2D positioning accuracy in terms of the Root Mean Square Error (RMSE) was about 0.40 m.