Iterated Maximum Mixture Correntropy Kalman Filter and Its Applications in Tracking and Navigation

In this article, the robust state estimation for discrete-time systems subject to non-Gaussian process and measurement noises (PMNs) is considered. Employing the ability of mixture correntropy in tackling non-Gaussian noise (NGN), we construct the cost function in accordance with the maximum mixture correntropy criterion (MMCC) based on the nonlinear regression model. The fixed-point iteration method is then employed to solve the cost function, and the measurement matrix is updated using the latest estimation. To enhance the estimation performance of the iterated estimation algorithm, the kernel parameters are adjusted adaptively in real time using the Mahalanobis distance (MD). The algorithm for the linear system is also provided. The performance of the proposed algorithms is verified through both simulations and experiments.