This file compares three different versions of the Kalman filter.

The Kalman filter is used for recursive parameter estimation.

The Kalman filter can handle noisy measurements.



The first implemented filter (fcn_KF) is the Kalman filter with standard

update of the covariance matrix P.

The covariance matrix reflects the uncertainties of the predictions.



To improve the numerical stability Potter developed a

square root update (fcn_KF_SRP) of the covariance matrix P.

Another version is the square root covariance update via

triangularization (fcn_KF_SRT).



This file generates a model. Then the three Kalman filters perform an

estimation of the model parameter. At the end the results are compared.



Sources:



Simon, D. (2006): Optimal state estimation



Kaminski, P. (1971): Discrete Square Root Filtering: A Survey of Current Techniques



Golub, G. (1996): Matrix Computations