Universal linearization concept for extended Kalman filters

The performance of a universal-linearization-concept-based extended Kalman filter (EKF) is evaluated by experimentally comparing its performance to that of a classical, linearization-based EKF, in the case of a simple nonlinear dynamical system. Instances of superior performance of the universal-linearization-based EKF are observed. In the case of nonlinear dynamics and linear measurements, the estimation advantage of the universal-linearization EKF increases when the process noise intensity decreases. Conversely, in the case of linear dynamics and nonlinear measurements, the estimation accuracy advantage increases when the process noise intensity increases. Furthermore, the universal-linearization EKF is more robust with respect to variations in the dynamics´ parameters, in both linear and nonlinear dynamics cases. The advantage of the universal-linearization EKF is more pronounced in the case of small process noise intensity