ABSTRACT
The most successful application of Kalman filtering (KF) has been for aerospace applications where considerable time and effort had been spent on generating accurate dynamic models and understanding the statistical nature of the measurement and process noise, the latter being the modelling of turbulence. The least squares (LS) and KF methods are based on minimizing the expected value of the variance of the estimation error, LS can be considered as a special norm or case of the minimum mean square error. Many forms of robust KFs are available, most are only modifications of existing KF solutions. The robustness means that the designed estimator is supposed to give estimates of the states of the dynamic system with acceptable and reasonable performance despite the presence of uncertainties that are not known a priori and not taken into account while building the structure of the filter.
