Tracking Filter Design for a Maneuvering Off-Road Ground Target

In this paper, two methods for tracking an off-road ground target using a horizontal coordinated turn motion model are proposed. The first method is the KF - KF method where the first linear Kalman filter estimates the target's turn rate and the second one uses the turn rate updated estimate from the previous filter to estimate the current position and velocity states of the target. The second method is a two model EKF/UKF based IMM filter, the motion models being nearly constant velocity and horizontal coordinated turn models. Both Extended and Unscented Kalman Filters are used in the second method and results compared for two sampling rates. Performance comparison is carried out by finding the root mean square values of the estimates for the two methods and both methods show good performance. The KF -KF has a low computation complexity over the IMM filter but unstable due to the divergence of error dynamics. Also, in method II, 0.2s sampling rate yields better outputs than 0.1s and the UKF based IMM shows better performance over EKF.