一种基于联邦卡尔曼滤波器的多源信息融合定位算法

A Multi-sensor Fusion Positioning Algorithm Based on Federated Kalman Filter

为提高车载组合导航定位系统的容错能力和信息源扩展能力,设计了一个联邦卡尔曼滤波器,搭建了一种开放式算法架构,选取了SINS/GNSS/里程计/高程计四种典型的车载定位信息源进行融合。算法框架以SINS为主参考系统,分别与其它信息源组成了子滤波器,子滤波器的输出结果经故障诊断和系统重构后,进入主滤波器进行信息融合。进行了仿真试验和实车试验,试验结果表明:该算法在降低误差状态维数,具备容错能力的情况下,达到了集中式卡尔曼滤波器的定位精度,提高了车载定位系统的环境适应性和信息源扩展性。

Focused on multi-sensor fusion for vehicle navigation in difficult environments,a federate Kalman filter is designed and an open algorithm architecture is presented.A integrated navigation system is build around SINS which is used as a core sensor and is consisted the sub-filters with other sensors(including GNSS,OD,Baro-altimeter).The sub-filters'output is fused in main filter after fault diagnosis and system reconfiguration.Simulation and experimental results are presented that the improved algorithm could achieve better positioning accuracy than the centralized Kalman filter in long time complicated environments error.State dimension could be reduced by using federated Kalman filter and best fault tolerance is achieved.Environmental adaptability and information expansibility of the vehicle positioning system are improved observably.