基于两阶段卡尔曼滤波的多传感器信息融合

Multi-sensor information fusion based on two-stage Kalman filter

在目标跟踪中,对目标运动建模时,常会遇到系统状态方程存在偏差问题.传统的信息融合方法总是假设系统状态方程中的偏差为常量,很少涉及偏差为随机变量的情形,但实际建模中常会出现这类问题.针对此问题,提出了基于两阶段卡尔曼滤波的多传感器信息融合方法.这种方法可以有效地消除系统状态方程在建模存在随机偏差时给信息融合所带来的影响,从而提高了融合精度.

We often meet with the problem when we construct the state models in the presence of random bias in target tracking.The traditional multi-sensor information fusion method is based on the assumption that the bias of state models is non-random.Little research is made into the problem of constructing the system models in the presence of random bias.Considering this problem,we describe an information fusion method based on two-stage state estimation in the presence of random bias for multi-sensor information fusion systems.This method can solve the problem effectively and increase the precision of fusion.