Online three-axis magnetometer calibration for a pedestrian navigation system using a foot-mounted inertial navigation system

Abstract

Foot-mounted inertial navigation systems(INS)having inexpensive micro electro mechanical system(MEMS)-based inertial sensors are widely used for pedestrian navigation.A foot-mounted INS can be combined with a magnetometer to constrain the heading angle error,but the magnetometer needs to be calibrated before use.This paper presents a magnetometer error model and an online calibration algorithm based on the foot-mounted INS characteristics.The magnetometer error characteristics and the foot-mounted INS mobility characteristics are used to develop a state equation and a magnetometer error measurement equation.An extended Kalman filter(EKF)is used for the online estimation and real-time calibration of the three-axis magnetometer errors with the zero velocity update(ZUPT)algorithm and a magnetic heading angle constraint algorithm for error constraint.The algorithm is validated by walking in a square playing ground.The results show that the online estimation and calibration algorithm reduce the end position error of the east direction from-110.7to 1.8mand the end position error of the north direction from 37.8to 5.2m compared to the system without calibration. This algorithm provides online calibration of magnetometer errors and significantly improves the positioning accuracy of pedestrian navigation.