New Tracking Filter Algorithm Using Input Parameter Estimation

A new method for the design of tracking filters for maneuvering targets, based on kinematic models and input signalsestimation, is developed. The input signal's level, u, is considered a continuous variable and consequently the input estimationproblem is posed as a purely parameter estimation problem. Moreover, the application of the new tracking filter algorithm isnot contingent on distinguishing maneuvering and non-maneuvering targets, and does not require the detection of maneuveronset. The filter will automatically detect the onset of a maneuver. Furthermore, an estimate of the target's acceleration isalso obtained with reasonable precision. This opens the door to the employment of advanced Augmented ProportionalNavigation Missile guidance laws, which require an estimate of the target's state acceleration, recognizing the precision is notas good as that for position and velocity estimates. When the target dynamics and measurement are linear and input u isconstant, then an unbiased estimate of of u and of the target state is obtained, provided that an observability condition holds.It is shown that the critical observability condition holds for kinematic target motion models of interest.