Adaptive filtering techniques for manoeuvring targets

A number of currently available maneuver compensation techniques are summarized. These techniques can generally be classified as parametric (the maneuver in modeled as a random event) or nonparametric (deterministic maneuver representation) approaches. The finite memory filter and input estimation procedure seemed to provide the most favorable combination of performance, computational requirements, and robustness. It is noted, however, that if the Markovian probabilities are accurately chosen, then the IMM (interactive multiple model) approach can be quite accurate. In situations where maneuvers are abrupt due to the pilot commands, the dynamic model based upon a pure jump process of the Poisson type yields promising results