基于交互式多模型因子图的自适应组合导航算法
Adaptive integrated navigation algorithm based on interacting multiple model factor graph optimization
曾庆化 1王守一 1李方东 1邵晨1
作者信息
- 1. 南京航空航天大学 自动化学院,南京 210016
- 折叠
摘要
针对复杂城市环境下因外部干扰或传感器故障而引起的传统车载导航系统定位精度下降的问题,提出了一种基于因子图的交互式多模型车载导航算法.基于因子图优化算法建立了IMU/GNSS/LIDAR组合导航系统模型,引入了交互式多模型对子系统传感器量测进行建模并构建变量节点,利用模型概率更新来优化传感器权重,并依据因子图非线性优化和增量平滑理论实现车载导航系统的解算与更新.实验结果表明:相比于自适应因子图算法,所提算法在复杂城市环境下的定位精度提高了 26.2%.
Abstract
To solve the problem of poor positioning accuracy of traditional vehicle navigation system caused by external interference or sensor failure in complex urban environments,an adaptive integrated navigation algorithm based on interactive multiple model factor graph optimization is proposed.The IMU/GNSS/LIDAR integrated navigation system model is constructed based on the factor graph optimization algorithm.The interactive multiple model is applied in the modeling process of sensor measurements and constructing variable nodes.The model update probability is used to optimize the sensor weights and the solution and update of the vehicle navigation system is realized based on the nonlinear optimization and incremental smoothing theory of factor graph algorithm.The experimental results show that compared with the adaptive factor graph optimization algorithm,the proposed algorithm can improve the positioning accuracy of vehicle navigation system in complex urban environments by 26.2%.
关键词
因子图/交互式多模型/车载导航/复杂场景Key words
factor graph/interacting multiple model/vehicle navigation/complex environment引用本文复制引用
基金项目
国家自然科学基金项目(61533008)
南京航空航天大学研究生科研与实践创新计划项目(xcxjh20220320)
出版年
2024
NETL
NSTL
维普
万方数据