An extended signal-correction observer (ES-CO) is presented for signals correction and estimation, which not only can reject the large measurement error, but also the system uncertainty can be estimated, in spite of the existence of intense stochastic non-Gaussian noise. Multi-input describing function method is proposed to analyze the ESCO robustness in frequency domain. The ESCOs are
developed for position/velocity and attitude angle/angular rate integrations, respectively, and they are applied to an aircraft navigation: Based on the ESCO, the position, attitude angle and the uncertainties in the flight dynamics are estimated. Experiments demonstrate the effectiveness of the proposed method.