On process noise covariance estimation

This paper proposes a method for estimating the process noise covariance matrix, using multiple Kalman filters. The basic idea is to employ the difference between the expected prediction error covariance, calculated in the Kalman filters, and the measured prediction error covariance. The required estimate of the process noise covariance is obtained by solving a least squares problem. One simulated example is used to illustrate the main benefits of the proposed method.