One of the most problematic factors in inertial/magnetic sensing-based tilt/azimuth estimation is magnetic distortion that can be added in magnetometer signals and may decrease the azimuth estimation accuracy. In this paper, we propose an orderswitching magnetic distortion compensation mechanism for accurate azimuth estimation. The proposed compensation mechanism switches between a first-order Gauss-Markov (GM) model and a second-order GM model of magnetic distortion. When the secondorder model is selected, the time-derivative of the magnetic distortion is additionally augmented in the states of the corresponding second-order azimuth Kalam filter. The proposed mechanism was experimentally validated in four different magnetically distorted conditions. The results show that the azimuth estimation error of the proposed method was less than 3° in all the tests and the proposed mechanism outperformed conventional approaches in the azimuth estimation.
Read full abstract