Adaptive sigma point filtering for state and parameter estimation

This article presents practical adaptive nonlinear filters for recursive estimation of the state and parameter of nonlinear systems with unknown noise statistics. The adaptive nonlinear filters combine adaptive estimation techniques for system noise statistics with the nonlinear filters that include the unscented Kalman filter and divided difference filter. The purpose of the integrated filters is to not only compensate for the nonlinearity effects neglected from linearization by utilizing nonlinear filters, but also to take into account the system modeling errors by adaptively estimating the noise statistics and unknown parameters. Simulation results indicate that the advantages of the adaptive filters make these attractive alternatives to the standard nonlinear filters for the state and unknown parameter estimation in the orbit determination.