一种基于AR运动模型CKF算法的无人机导航

摘要

Filtering algorithms have been widely used in GPS/INS integration based UAV navigation.The motion model,as the key component in the filter prediction,is critical for the algorithm performance. However,the traditional motion model only uses a fixed motion mode for the prediction,which is not able to describe the UAV motion accurately.In this paper,an AR motion model based CKF algorithm for the UAV state estimation is proposed.By constructing the real-time dynamic updating AR model with sliding window,the adaptive filtering is designed for the UAV states estimation.The simulation results show that the positioning estimations from the designed CKF-AR based filtering algorithm are superior than the other traditional motion model based filtering results.%针对无人机组合导航滤波算法中传统的运动模型单一固定,灵活性差,无法精确描述无人机复杂多变的运动状态,甚至还会导致滤波发散等问题,提出了一种基于AR运动模型的CKF算法,并应用于无人机导航定位中.通过滑动窗的方法,构建实时动态更新AR模型估计物体运动状态,并且结合CKF进行滤波,从而有效提高无人机导航性能.仿真实验结果表明,该算法能够有效提高无人机导航定位精度,优于其余几种基于传统运动模型的滤波算法.