Abstract In order to improve the quality of sampling,the use of adaptive Unscented Kalman filtering (AUKF) for importance sampling in particle filter is researched as the standard particle filter has degeneracy. The adaptive factor based on adaptive filtering is used to adjust the proportion of observation and dynamic model of Unscented Kalman filtering (UKF), and to make the covariance of the predicated vector approach the true value. This way,it can improve the accuracy of importance sampling, and further improve the accuracy of particle filters. The precision of particle filter using adaptive UKF is the best among the improved particle filter by use of sequential importance sampling (SIS), extended Kalman filtering (EKF), adaptive extended Kalman filtering (AEKF), Unscented Kalman filtering (UKF) and adaptive Unscented Kalman filtering (AUKF) for importance sampling in a simulated example. It has been proveed that particle filter taking adaptive UKF as importance sampling function is a kind of efficient way to improve the precision of particle filter.
