State switch UKF algorithm and its application in attitude estimation

Considering the sharply increased computation burden caused by the increasement of state dimension multiplication in Augmented unscented Kalman filter (AUKF), a state switch unscented Kalman filter (SUKF) algorithm whose process and measurement noise has no correlation was proposed. Via selecting different state variables in time update and measurement update stage, the state dimension of real-time filtering was reduced and the number of Sigma points selected decreased, so the filter computation was reduced and speed was improved. Aming at the quaternion normalization restrictions in attitude determination, a parameter-switching algorithm was proposed. Through switching the quaternion parameter and the modified rodriguez parameter (MRPs), the quaternion weighted meanand covariance singularity problems were solved. The SINS/CCD attitude simulation results show that: compared with the AUKF, the proposed algorithm estimation accuracy is about equal and the estimation time reduces approximately 1/3.