For a class of uncertain systems with large-error sensing, the low-order stable signal corrector and observer are presented for signal correction and uncertainty estimation according to completely decoupling estimation. The model-free signal corrector can reject the bounded stochastic disturbance/error in global position sensing, and system uncertainty can be estimated by the observer, even the existence of large disturbance in position sensing. Furthermore, a general form of signal corrector is given. The describing function method is used to analyse the robustness of the corrector in frequency domain, and the parameter selection rules are presented. The merits of the signal corrector includes its model free, gain-bounded stable structure, sufficient rejection of bounded stochastic disturbance/error in sensing and ease of parameters’ regulation. The corrector and observer are applied to a UAV navigation and control for large disturbance/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, experiments demonstrate the effectiveness of the proposed method.