基于KALMAN滤波的两种多传感器观测融合方法

目前有两种基于K a lm an滤波的多传感器观测融合方法,方法1是将观测向量的维数增加,获得扩展观测向量。方法2是在最小均方误差准则下,对不同传感器间的观测向量进行加权运算,获得与单个传感器相同维数的观测向量。通过对滤波器的状态估计误差协方差的分析和相关数学表达式,给出了两种方法的对比。仿真结果表明,当两个传感器的观测矩阵相同时,两种方法在功能上等价,但方法2的运算复杂度低。当两个传感器的观测矩阵的维数相同,但其值不相等时,方法1优于方法2。当两个传感器的观测矩阵的维数不同时,只能用方法1,而方法2失效。

Two Measurement Fusion Methods based on Kalman Filtering for Multisensor Data Fusion

Currently there exist two commonly used measurement fusion methods for kalman-filter-based multi-sensor data fusion.The method one simply obtains the expanded-observation vector through increasing the number of dimension of the observation vector.The method two obtains the observation vector,which has same dimensions with the single-sensor,through calculation of the coefficient of the power between different sensors based on a minimum-mean-square-error criterion.The paper draws comparisons with the two methods based on an analysis of the fused state estimate error and covariance.The result of simulation shows that the two fusion methods are functionally equivalent,while the two need a low-complex calculation,if they have the same observation matrices.If they have the same dimensions but not the same values,the one is outbalance the two.