An Improved Adaptive and Robust Initial Alignment Method for Rotation MEMS-based SINS

Abstract

This paper proposes an adaptively robust unscented Kalman filter (ARUKF) for the rotation micro-electro-mechanical system based strapdown inertial navigation system (MINS) to achieve fast in-motion initial alignment in the presence of large misalignment angles. First, UKF is utilized to address nonlinearity issues resulting from large misalignment angles. Second, the strong tracking strategy is implemented to robustly compensate for dynamic model errors during the transition phase. The variational Bayesian is then applied in the steady state to adaptively estimate the time-varying measurement noises. The proposed method speeds up convergence during the transition phase and improves convergence precision during the steady phase. In conclusion, the turntable experiments verify the validity of the proposed method.