Unscented Kalman Filtering for SINS Attitude Estimation

This paper presents a nonlinear error model based on the quaternion for attitude estimation of the strapdown inertial navigation system (SINS). Extended Kalman filter (EKF) is widely applied in the attitude estimation problem of SINS. The unscented Kalman filter (UKF) is an extension of the classic EKF to nonlinear process and measurement models. It is noted that for the nonlinear system UKF uses a carefully selected set of sample points to map the probability distribution more accurately than the linearization of the standard EKF. Then, unscented attitude filter is designed to achieve the nonlinear filter based on the proposed model. The attitude kinematics error model is described by a quaternion because no singularities are present and kinematics equation is bilinear. Monte Carlo simulation is made to compare the new filter with the standard EKF. The results and analysis indicate that UKF has the faster convergence rate, the higher filtering accuracy, and more stable estimation performance in attitude estimation of SINS.