A robust adaptively filtering algorithm in GPS/DR integrated navigation

Abstract

For the mathematical model based on GPS/DR Vehicle Integrated System is non-linear and it's linear filter arouse much errors by EKF, we introduce adaptively robust filter. The simulative computation of the adaptively robust filtering and adaptively kalman filtering are studied respectively in this system. The results show that robust adaptive filtering can determine the covariance matrix of observation noise adaptively and regulate covariance matrix of state parameter noise by adaptive factor. This method can control the impacts on the valuation of state parameters aroused by Observed noise abnormal and dynamic model noise abnormal and. make valuation more reasonable. The experimental result shows that adaptively robust filtering algorithm can outperform the adaptively kalman filtering algorithm in terms of accuracy.