Gaussian-sum cubature Kalman smoothers for bearings-only tracking

Abstract

In this paper, a fixed-lag and a fixed-interval Gaussian-sum cubature Kalman smoother are proposed for the bearings-only tracking problem. The smoothers are of the forward-backward type and they utilise the Gaussian-sum cubature Kalman filter with improved robustness presented by the authors in [1]. Simulation results show that both the fixed-lag and fixed-interval smoothers exhibit improved accuracy over their filtering counterpart and outperform other existing smoothers of the same type for this problem, with the root-mean-square error overlapping the Cramér-Rao lower bound.