抄録
In micro satellites, the robustness and calculation cost of attitude estimation algorithm are the most importance ones because micro satellites sometime use low cost, low accuracy attitude sensors and low powerful onboard computer. In order to improve these qualities of attitude determination system (ADS), the advance estimation algorithms will be integrated in on-board computer of satellites. The most well-known and usefull algorithms used for this nonlinear estimation are the Extended Kalman Filter (EKF) and the Unscented Kalman Filter (UKF). The standard EKF makes use of the first-order Taylor series to approximate the nonlinear state and measurement functions. The implementation of EKF is fast however EKF is sensitive with the selection of process noise, measurement noise matrixes and the initial error. On the other hand, the UKF uses a carefully selected set of sample points called sigma points to accurately map the probability distribution of the state vector. UKF is more robust than EKF but the calculation cost takes about 10 times higher than EKF. This paper presents a new method based on the mixed of EKF and UKF algorithms, this method has same accuracy and robustness level with UKF but the calculation cost drops below half of UKF. Several numerical simulation examples are given to verify the proposed algorithm and to show its improved performance.
