W. Dale Blair
Georgia Tech Research Institute
Although the Kalman filter has been widely applied to target tracking applications nearly since its introduction in the early 1960’s, until recently, no systematic design methodology has been available to predict maneuvering target tracking performance or to optimize filter parameter selection. This lecture, presents a rigorous procedure for selecting optimal process noise variances for the Kalman filter based on a particular sensor and target model. Design of nearly constant velocity (NCV) Kalman filters with discrete white noise acceleration and exponentially-correlated acceleration errors are addressed. Furthermore, filter design for tracking maneuvering targets with linear frequency modulated waveforms is considered as well.