Attitude determination using gyros and vector measurements aided with adaptive kinematics modeling

Abstract

High-precision attitude estimation in real time plays a vital role in spacecraft flight. In this work, the attitude filtering using gyros and vector measurements is studied in the framework of the traditional multiplicative extended Kalman filtering (MEKF). An adaptive online kinematic modeling is proposed to aid the filtering, resulting in the so called adaptive kinematics-aided Kalman filtering (AKKF). Meanwhile, the state estimate is reset to be zeros and the covariance resetting is implemented accordingly at each step. Due to time-varying features of the attitude motion, a predetermined statistical model for the employed kinematic model cannot generally reflect the reality. So, the best invariant quadratic unbiased estimates (BIQUE) is adopted to adaptively tune the variance components in the employed kinematic model, to better model the realistic kinematic characteristics and hence to further improve the accuracy. Numerical simulation study indicates the reliability and accuracy of the proposed AKKF method.