A Kalman tracker with a turning acceleration estimator

Abstract

In tracking systems using phased array antennas, the simple Kalman filter has been employed due to the development of high-speed computers in recent years. The Kalman filter performs almost perfect tracking when a target takes a straight course. However, the quality of the Kalman filter degrades significantly for maneuvering targets since the statistics of acceleration are assumed to be a white Gaussian process in the filter. The Kalman tracker with a simple input estimator and the two-stage Kalman estimator have been proposed for maneuvering target tracking that uses the residuals to estimate the target acceleration and that requires no a priori information of the magnitude of the maneuver. However, the estimate of the acceleration might not be accurate in the case that the residues increase dramatically such as the circular turn. In this paper, a Kalman tracker with a turning acceleration estimator is proposed that estimates the turning acceleration directly from the feature of the maneuvering target trajectory. A computer simulation is performed for various target profiles to show the effectiveness of the proposed tracker as compared with conventional trackers. © 2000 Scripta Technica, Electron Comm Jpn Pt 1, 84(1): 1–11, 2001