An adaptive Gaussian sum approach for maneuver tracking

A technique for tracking a target through a maneuver that adjusts the motion model based on the current measurement information to detect the maneuver is explored. Modeling of the target motion uses an adaptive function approximation technique based upon the concept of Gaussian sum function approximation. The parameters that describe each Gaussian are identified using a Kalman filter in such a way as to emulate the mathematical function that represents the target motion or the error between the mathematical model and the true target dynamics. The incorporation of the Gaussian sum into the track estimator results in a coupled Kalman filter. As a result of this coupling, this filter simultaneously estimates both the states of the target track and the parameters of the Gaussian sum. This improves the motion model for the maneuver and results in a better prediction of the target track which in turn enhances the estimates of the updated state