Numerically well conditioned implementations of the PDA and PMA filters

New implementations of the standard Kalman filter have been developed by a number of researchers in order to minimize the undesirable effects of using finite-wordlength arithmetic. A number of extensions to the Kalman filter have also been developed for handling estimation problems with measurement and input uncertainty. It has therefore become necessary to develop implementations of these filters which are less susceptible to finite-wordlength effects. The authors introduce implementations of filters that deal with measurement and input uncertainty that are less susceptible to the effects of using finite-wordlength arithmetic than the original filters