Adaptive Maximum Correntropy Based Robust Cubature Kalman Filter Under Non-Gaussian Noise

This paper introduces a novel adaptive Maximum Correntropy Criterion (MCC)-based robust cubature Kalman filter (AMCRCKF) designed to improve state estimation accuracy and prevent filter divergence under conditions such as non-Gaussian and impulsive noise, signal loss, or outliers. The algorithm enhances the existing CKF by incorporating a fading factor and the MCC in its cost function to effectively manage outliers. An adaptive factor within the Kalman Filter framework optimizes the balance between kinematic model forecasts and measurement updates. Simulation results confirm that the algorithm significantly boosts the robustness and adaptability of the traditional CKF, showcasing its superior performance in handling model inaccuracies and measurement outliers.