Gaussian Mixture Measurement for Very Long Range Tracking

Abstract

Target tracking with very long range is studied in this paper. Such tracking problem has severe measurement nonlinearity that will cause consistency problems and large tracking errors. Gaussian mixture measurements are obtained by dividing the measurement likelihood into several Gaussian components. The Gaussian Mixture Measurement-Integrated Track Splitting (GMM-ITS) is applied to very long range tracking scenarios. The simulation results show that the GMM-ITS can produce consistency in the filtering results crucial to the filter performance. Furthermore, it is also able to estimate the target state accurately with small tracking errors.