Abstract:

Considering the divergence phenomenon of extenthed Kalman filter (EKF) in SINS/GPS tightly integrated navigation system, a square root unscented Kalman filter (QSRUKF) algorithm based on quaternion  is proposed. In order to ensure normalization of unit quaternion in SRUKF algorithm, an attitude matrix cost function is established to solve the problem of predicted average quaternion by minimizing the cost function; the multiplicative quaternion error is used to predict covariance, since it represents the distance between predicted quaternion and mean quaternion. On the basis, the specific steps of QSRUKF in SINS/GPS integrated navigation system are given. Simulation results show that the proposed algorithm has higher precision and stronger robustness under the larger initial attitude error angle compared with the EKF algorithm.