Hybrid Unscented Kalman Particle Filter for Ground Target Tracking

We propose to apply two nonlinear filtering techniques, namely, the unscented Kalman filter and the particle filter, to the tracking of multiple ground targets in clutter. Both the target dynamics and radar measurements are highly nonlinear. Instead of truncating the nonlinear functions to the first order as in most of current implementations with the extended Kalman filter, the proposed unscented Kalman filter and the particle filter approximate the distribution of the state with a finite set of points (i.e., the particles) and then propagate these particles through the true nonlinear functions. Because the nonlinear functions are used without approximation, it is much simpler to implement and can generate better result. The proposed hybrid ground target tracking algorithm will integrate these two nonlinear filters with an iterated probabilistic data association technique to handle clutter, an adaptive multiple model estimator to deal with target maneuvers, and pseudo measurement and state constraints to incorporate road network and topographic information. In Phase I, the proposed hybrid ground target tracking algorithm will be formulated with the computational algorithms developed. We will use a computer simulation program to evaluate the algorithms. In Phase II, the recommended architecture and the validated algorithms will be implemented to process simulated and/or real data for demonstration.