基于鲁棒AFCKF的无人车多源组合导航方法  

Multi-source integrated navigation method for unmanned ground vehicles based on robust AFCKF

作　　者：宋迎春[1,2] 冯抗洪 崔先强 SONG Yingchun;FENG Kanghong;CUI Xianqiang(School of Geosciences and Info Physics,Central South University,Changsha 410083,China;School of Resources and Environment,Yili Normal University,Yining 835000,China)

机构地区：[1]中南大学地球科学与信息物理学院,长沙410083 [2]伊犁师范大学资源与环境学院,伊宁835000

出　　处：《中国惯性技术学报》2023年第12期1203-1209,1219,共8页Journal of Chinese Inertial Technology

基　　金：国家自然科学基金(42174040,41674012)。

摘　　要：针对无人车进行多源信息融合时部分传感器失效或量测异常导致定位精度下降甚至无法定位的问题,提出了一种鲁棒自适应联邦容积卡尔曼滤波(AFCKF)算法。首先,在子滤波器中将Huber方法与容积卡尔曼滤波相结合,并基于马氏距离实时调整Huber方法中的调节因子,提高了子滤波器的估计精度和鲁棒性。其次,在信息融合中,基于预测状态残差以及量测残差的二次型引入一种自适应信息分享因子计算方法,减小了子滤波器不准确估计对主滤波器信息融合的影响。最后,对所提方法进行了仿真实验。仿真实验结果表明,相比于传统的联邦容积卡尔曼滤波和基于Huber方法的联邦容积卡尔曼滤波,所提方法的平均定位精度分别提高了77.84%和17.20%,平均速度精度分别提高了63.25%和9.30%。In order to address the problem of decreased positioning accuracy or even the inability to locate in unmanned ground vehicles during multi-source information fusion,caused by sensor failures or measurement anomalies,a robust adaptive federated cubature Kalman filtering(AFCKF)algorithm is proposed.Firstly,the Huber method is combined with the cubature Kalman filter in the sub-filters,and the tuning factor in the Huber method is adjusted in real time based on the Mahalanobis distance to improve the estimation accuracy and robustness of the sub-filters.Secondly,in information fusion,an adaptive method for calculating information sharing factors has been introduced based on the quadratic form of predicted state residuals and measurement residuals,which can dynamically reduce the impact of inaccurate estimates from sub-filters on the information fusion of the main filter.Finally,the simulation experiments of the proposed method are carried out.Simulation experimental results show that compared to traditional federated cubature Kalman filtering and federated cubature Kalman filtering based on the Huber method,the average positioning accuracy of the proposed method is improved by 77.84%and 17.20%,and the average velocity accuracy is improved by 63.25%and 9.30%,respectively.

关 键 词：多源组合导航 联邦容积卡尔曼滤波 鲁棒估计 自适应信息分享因子 

分 类 号：TN967.2[电子电信—信号与信息处理]