Robust Kalman filtering with generalized Gaussian measurement noise
W. Niehsen · IEEE Transactions on Aerospace and Electronic Systems · 2002
A recursive state estimator based on adaptive generalized Gaussian approximation of the innovations sequence probability density function is constructed. The proposed state estimator is computationally efficient and robust in the case of heavy-tailed measurement noise. Compared with standard Kalman filtering, significant improvements with respect to stationary mean square error and rate of convergence are achieved.