Correlated System Noise Error Tracking Filter Design for a Sharply Maneuvering Ground Target

This paper is about the design of a correlated system noise error tracking filter for a sharply maneuvering off-road ground target. A high process noise Unscented Kalman Filter based two-model Interacting Multiple Model filter is designed to match the evasive target conditions. In addition to uncertainties in the system dynamics, the terrain profile is assumed to be part of the input noise to the system. The overall noises in position and velocity are taken to be correlated and assumed to be generated by a linear shaping filter. The Kalman filter for the discrete white noise acceleration model and the unscented Kalman filter for the highly nonlinear horizontal coordinated turn model with unknown turn rates are designed. The simulated results show good filter performance with low mean square errors of target position and velocity. The filter also shows good stability with nondivergent error dynamics. The computational challenge of the designed filter is observed at the maneuver entry and exit but despite severe target maneuvers, the filter is able to maintain its tracks.