Abstract: The integrated navigation technology can contribute greatly to solving land mobile robot autonomous navigation,in which the GPS/DR is one typical system.In view of the non-linear characteristic of the GPS/DR state equations,an improved federated UKF based on Unscented Kalman Filtering（UKF） and distributed information fusion technology is designed and used in the navigation system.Numerical simulation results indicated that,federated UKF has the better filter precision compared to federated EKF,simultaneously has a higher stability and the fault tolerance,is an ideal nonlinear filter method for GPS/DR navigation.

摘要： 组合导航技术是解决地面机器人自主导航的一个有效途径,其中GPS/DR是一种典型的组合方式。常用的卡尔曼滤波主要用于处理线性问题,针对该导航系统非线性的特点,对Unscented卡尔曼滤波（UKF）与分散式滤波技术相结合的方法进行了研究,建立了用于GPS/DR导航系统的联邦UKF算法。数值仿真实验表明,联邦UKF比联邦EKF有更好的滤波精度,同时有更高的稳定性和容错性,是一种理想的GPS/DR导航非线性滤波方法。