An improved adaptive Kalman filter algorithm is presented to model error and measurement uncertainty. Byusing the theory of linear matrix inequalities, the adaptive algorithm for model error is obtained by using an upperbound for the state prediction covariance matrix. The measurement uncertainty is solved using the idea of biascharacterization filter, which improves covariance fidelity in the presence of unknown measurement biases. Theproposed adaptive filter algorithm was successfully implemented in ISL based autonomous navigation for GNSS.Software simulation results indicated that the proposed adaptive filter provide promising performance in robustnessand accuracy compared with previous adaptive algorithms.
展开▼
