For the synthesis problem for an invariant tracking system for a nonlinear automated
control object under incomplete measurements, we develop a decompositional procedure of observer
state synthesis with sigmoidal correcting influences in order to get current estimates of
the unmeasured state variables and existing uncertainties. This observer in the pre-limit situation
possesses the advantages of an observer with discontinuous correcting influences operating
in sliding mode; in particular, it lets us estimate external influences without introducing their
dynamical model. Unlike a sliding mode observer whose order has been extended due to filters
over discontinuous correcting influences, the dimension of this observer equals the dimension of
the control object, and in a microprocessor implementation this observer ensures better quality
(smoothness) of the signals being estimated.