A novel robust two-stage extended Kalman filter for bearings-only maneuvering target tracking

Abstract

A novel robust filter is proposed via embedding extended Kalman filter into an optimal two-stage Kalman filter, which is improved through a modified Sage-Husa noise statistics estimator. Meanwhile, strong tracking multiple fading factors are introduced in this paper to improve the tracking performance for high maneuvering targets. The new method provides an optimal estimation of the target state through a combination of the output of the first stage (a “bias-free” filter) and that of the second stage (a “bias-compensating” filter). Furthermore, the unknown statistical parameters of virtual noises are estimated online, and the predicted covariance can be adjusted in real time by fading factors when high maneuvers occur. Simulation results of tracking a maneuvering target by 3 passive sensors show that the proposed algorithm has advantages over the conventional method in terms of the numerical accuracy with only a little additional computational cost. Key words: Target tracking, two-stage Kalman filter, noise statistics estimator, multiple fading factors.