Robust Unscented Quaternion Kalman Filter for Polar In-Motion Alignment and Virtual Polar Region Technology

Initial alignment is an indispensable technique for strapdown inertial navigation system (SINS) because the navigation accuracy of SINS depends on the performance of the initial alignment method. However, considering the rapid convergence of meridians and the small horizontal component of the Earth’s rotation in the polar region, traditional stationary base alignment and self-alignment methods cannot be achieved. In this article, an aiding global positioning system (GPS) is used to achieve the in-motion alignment process in transverse frame. Meanwhile, due to the existence of the multipath effect in the polar region, the measurement noise of GPS often exhibits outliers, which will degrade the initial alignment accuracy. Thus, a robust unscented quaternion Kalman filter (KF) algorithm under transverse frame is proposed for GPS-aided low-cost SINS in-motion alignment in this article, which can not only eliminate the outliers in GPS, but also effectively suppress the influence of gyro bias. After that, a virtual polar region technology based on virtual sphere model with transverse attitude, velocity, and height information invariable is proposed to overcome the difficulty in polar data acquisition. Simulation and field test results demonstrate the effectiveness of the proposed polar in-motion alignment algorithm and virtual polar region technology in transverse frame.