Pseudo Open-Loop Unscented Quaternion Estimator for Attitude Estimation

In this paper, a novel attitude estimator called pseudo open-loop unscented quaternion estimator (PO-USQUE) is proposed for attitude estimation. In PO-USQUE, the attitude error estimate is used to construct the sigma points in the time update of the next filtering recursion, and therefore, the reset operation in current filtering recursion is no longer needed. The attitude update is achieved through the construction of the quaternion-based sigma points. Compared with the traditional USQUE, the attitude in PO-USQUE is updated more accurately due to the special structure, and the corresponding covariance is updated simultaneously in a nature way. The performance of the proposed PO-USQUE is assessed through two representative attitude estimation applications using simulation and field experiment data. The superiority of the PO-USQUE over USQUE in terms of accuracy, consistency, and efficiency is validated by the simulation and test results.