Ped-Mag-ODO: Indoor Pedestrian Motion Speed Estimation Method Based on Dual Magnetometers

In the field of indoor pedestrian localization, accurate speed measurement is crucial for achieving continuous and reliable positioning of pedestrians. To address the limitations of existing indoor pedestrian speed measurement methods, a posture-immune dual magnetometers odometry speed measurement method (Ped-Mag-ODO) is proposed, which aims at enhancing the robustness of indoor localization. This method employs three-axis magnetometers installed at the forward and backward of a helmet to measure the surrounding magnetic field information, estimating speed by matching the magnetic field waveforms sampled from the magnetometers. The algorithm consists of three stages: 1) data preprocessing; 2) matching and velocity estimation; and 3) elimination of mismatch speeds. In the data preprocessing stage, three methods to process the raw three-axis magnetic field data are proposed: 1) gradient magnitude extraction; 2) continuous posture projection; and 3) stationary posture projection. These methods are designed to eliminate the effects of magnetometer bias and posture variations on the waveforms from the front and rear magnetometers, ensuring the consistency of the waveforms to be matched. In the matching and velocity estimation stage, the Double-Window Weighted Dynamic Time Warping algorithm is proposed to improve matching accuracy. In elimination of mismatch speeds stage, a “ Two-Step Mismatch Speed Elimination Algorithm” is introduced, that effectively identifies and eliminates outliers in the speed data, further enhancing the accuracy of speed estimation. Experimental results demonstrate that in indoor environments, whether pedestrians are walking or trotting, the proposed method exhibits performance comparable to that of a laser-inertial odometry (RMSE < 0.15 m/s).