An Adaptive UKF Filtering Algorithm for GPS Position Estimation

The unscented Kalman filter (UKF) is widely applied to different kinds of nonlinear filtering problems. But the performance of the conventional UKF algorithm is unstable because the fixed covariance parameters cannot accord with the vary situation. This paper proposes a new adaptive UKF filtering algorithm for the GPS based position estimation problem. In terms of the GPS system error characters, the new algorithm builds a model of the propagation error, and estimates its covariance by real time. The real satellite data were used to verify the algorithm then. The result of the experiment shows that the accuracy of new algorithm is better than the conventional UKF algorithm.