Rauch-Tung-Striebel high-degree cubature Kalman smoother

Abstract

In this paper, a new Rauch-Tung-Striebel type of nonlinear smoothing method is proposed based on a class of high-degree cubature rules. This high-degree cubature Kalman smoother generalizes the conventional third-degree cubature Kalman smoother and considerably improves its estimation accuracy. A target tracking problem is utilized to demonstrate the performance of this new smoother and compare it with other Gaussian approximation smoothers. It will be shown that the high-degree cubabure Kalman smoother outperforms the extended Kalman smoother, the unscented Kalman smoother, the third-degree cubature Kalman smoother, and maintains close performance to the Gauss-Hermite quadrature smoother with much less computational cost.