Desensitized cubature Kalman filter with uncertain parameters

In this study, a robust desensitized cubature Kalman filtering (DCKF) is proposed for nonlinear systems with uncertain parameters. Unlike the cubature Kalman filtering, the desensitized cost function is introduced by penalizing the posterior covariance trace with a weighted sum of the posteriori sensitivities. The sensitivity of the root square matrix is obtained by solving a Lyapunov-like linear equation, and the sensitivity propagation of the state estimate errors is presented. The effectiveness of the proposed DCKF is demonstrated by two numerical examples in which models with uncertain parameters are considered.