For a class of uncertain systems with large-error sensing, according to completely decoupling estimation, the nonlinear low-order stable signal corrector is presented for signal correction, and the extended state observer is used for uncertainty estimation. The signal corrector can reject the large error in global position sensing even the existence of stochastic non-Gaussian noise. In order to be fit for the hardware computational environments in engineering practice and to select the parameters more easily, the linear version of the low-order stable signal corrector is also developed, and the error transfer method is presented to prove it. The correctors and observers are applied to a UAV navigation and control for large-error corrections in position/attitude angle and the uncertainties estimation in the UAV flight dynamics. The control laws are designed according to the correction-estimation results. Finally, the experiments demonstrate the effectiveness of the proposed method.