Error Calibration of Magnetometer Using Nonlinear Integrated Filter Model With Inertial Sensors

Abstract

This paper presents an onboard heading estimation algorithm using IMU and strapdown magnetometer without other external heading references. To calibrate the magnetic deviation, sensor errors caused by hard iron effect and initial heading of strapdown magnetometers are considered. In our approach, sensor output distortion due to the soft iron effect is ignored, which is relatively small. First, for the estimation of heading angle, system and measurement model is presented. Then particle filter and extended Kalman filter is introduced for performance comparison. The proposed algorithm for the integration of IMU and magnetometer is verified via numerical simulation using Matlab. Simulation result demonstrates accurate heading estimation error under 1 degree for both algorithms when there exists a small initial heading error and hard iron effect, yet particle filter provides more robust and accurate result than the extended Kalman filter in case the initial heading error and biases are large.