Topography-based VS-IMM estimator for large-scale ground target tracking

The Interacting Multiple Model (IMM) estimator has been shown to be very effective in various target tracking problems. It is possible to vary the set of models in the IMM estimator based on some criteria to yield better estimates. This results in a Variable Structure IMM (VS-IMM) estimator where the mode set not only differs across targets, but also varies with time for a given target. This paper addresses the problem of tracking convoys of ground targets using Moving Target Indicator (MTI) reports to illustrate the development, operation and the benefits of an adaptive VS-IMM estimator. The targets under track are moving along a constrained path, for example, a highway, with varying obscuration due to changing terrain conditions. Here we present a VS-IMM estimator, where filter modules are adaptively modified, added or removed depending on the terrain topography, for tracking on-road and off-road targets within the same framework. At each scan, the structure of the estimator for every target is individually modified based on the known topography of the surveillance region and the predicted location of the target. This enables the estimator to handle the variation in the possible motion modes across targets as well as with time for each target. Because of the nonlinear nature of the MTI reports, which consist of two-dimensional positions and range rate, extended Kalman filters are used as IMM estimator modules.