鲁棒四元数平方根容积卡尔曼滤波算法  

Robust Quaternion Square-root Cubature Kalman Filter

作　　者：李开龙[1] 胡柏青[1] 常路宾[1] 李杨[1] 

机构地区：[1]海军工程大学导航工程系,湖北武汉430033

出　　处：《压电与声光》2015年第5期873-876,882,共5页Piezoelectrics & Acoustooptics

基　　金：国家"九七三"基金资助项目(2012CB719902);国家自然科学基金资助项目(61304241;61374206;41374082)

摘　　要：针对姿态估计领域中,当系统观测噪声不完全服从于高斯分布或观测量中含有野值时,传统的四元数卡尔曼滤波算法会降阶,使滤波估计性能降低等问题,提出了一种鲁棒化的四元数平方根容积卡尔曼滤波算法。通过研究引入M估计的Huber鲁棒化滤波框架,对四元数平方根容积卡尔曼滤波算法的量测更新进行修正,增强滤波估计的鲁棒性。以SINS/GPS位置松组合为应用研究背景,在不同仿真环境下,对比验证所提出算法对于姿态、速度和位置估计效果,试验结果表明,所提出算法具有更好滤波鲁棒性和稳定性。In order to address the problem that the conventional quaternion Kalman filter algorithm can be severely degraded when the distribution of the measurement noise deviates from the assumed Gaussian distribution or the observations are contaminated with outliers in the attitude estimation field,a novel robust quaternion square-root cubature Kalman filter based on Huber′s M-estimation is proposed.Through investigating the Huber-based robust filtering framework in which the M-estimation is introduced and the measurement update of the quaternion squareroot cubature Kalman filter is modified,the robustness of the filtering estimation is enhanced.The two quaternion Kalman filters are evaluated under different simulated environments for integrated SINS/GPS loosely coupled system.The results show that the proposed filtering algorithm in this paper has better robustness and stability over the conventional filtering algorithm.

关 键 词：姿态估计 卡尔曼滤波 M估计 四元数 鲁棒 

分 类 号：TN911.72[电子电信—通信与信息系统]