State Estimation under Outliers by the Maximum Correntropy Extended Kalman Filter

In this study, we treat the problem of state estimation under measurement outliers for nonlinear systems. The Kalman filter (KF) is optimal under linear systems with Gaussian noise but will deteriorate seriously if outliers such as impulse noises are involved. In this work, we propose a nonlinear system state estimator based on the extended Kalman filter (EKF) with the maximum correntropy criterion (MCC), and improve the MCC with a fixed-point iteration. With the novel MCC as the weight of Kalman gain, the EKF can counteract the influence of outliers during estimation. A series of simulation results shows that the maximum correntropy extended Kalman filter (MCEKF) has much better performance than the traditional EKF under outliers.