Novel robust generalized high-degree cubature kalman filter for transfer alignment

A novel robust generalized high-degree Cubature Kalman filter (RGHCKF) for transfer alignment is proposed to solve the problem that the Cubature Kalman filter (CKF) declines in accuracy and even diverges when the non-Gaussian noise and outliers exist in the observation. In RGHCKF, in order to achieve higher accuracy, the generalized high-degree CKF (GHCKF) time update is carried out and the innovation chi-square test method is used to prejudge the degree of contaminative observation, and the observation is modified by using the Huber method, and then the GHCKF measurement update is implemented, so that the robustness is realized. The proposed algorithm doesn’t need approximating nonlinear measurements model by using the statistical linear regression model. Simulations and vehicle tests show that the RGHCKF has superior performance in robustness and estimation precision.