基于精化预积分的GNSS/IMU/视觉多源融合定位方法  

A GNSS/IMU/vision multi-source fusion localization method based on refined pre-integration

作　　者：贾晓雪 赵冬青[1] 肖国锐[1] 杨显赐 杨朝永 赖路广 JIA Xiaoxue;ZHAO Dongqing;XIAO Guorui;YANG Xianci;YANG Chaoyong;LAI Luguang(Institute of Geospatial Information,Information Engineering University,Zhengzhou 450001,China)

机构地区：[1]信息工程大学地理空间信息学院,郑州450001

出　　处：《北京航空航天大学学报》2024年第6期2026-2032,共7页Journal of Beijing University of Aeronautics and Astronautics

基　　金：国家自然科学基金(41774037,41904039,42274045)。

摘　　要：针对传统预积分算法固定地球重力值和忽略地球自转的问题,提出一种考虑地球自转和重力变化的惯性测量单元(inertial measurement unit, IMU)预积分算法。参照高精度捷联惯性导航解算的动力学模型,在IMU预积分动力学模型的姿态更新中引入地球自转角速率,速度和位置更新中引入由地球自转引起的科里奥利加速度,同时将由载体位置引起的地球重力变化及时反馈至预积分算法中,详细推导了引入地球自转和重力变化后预积分算法的具体过程,实现对传统预积分模型的精化。并将精化的预积分算法应用于基于紧耦合全球导航卫星系统(global navigation satellite system, GNSS)/IMU/视觉多源融合系统中,实测实验结果表明:利用精化的预积分模型可使系统预积分的模型误差有效减小,显著提升多源融合系统整体的定位定姿精度,其中系统定位精度提升32.41%,航向角精度提升4.23%。A pre-integration algorithm taking into account the earth's rotation and gravity change is developed in order to address the issue that the conventional pre-integration algorithm fixes the value of the earth's gravity and ignores the Earth's rotation.Referring to the dynamic model of high-precision strapdown inertial navigation,the earth rotation angle rate is introduced in the attitude update of the pre-integral dynamic model,and the Coriolis acceleration caused by the earth rotation is introduced in the velocity and position update.Simultaneously,the traditional pre-integral model is improved,taking into account that the carrier's position can feed back the gravity change to the pre-integral algorithm in time.All the process of the pre-integral algorithm are derived in detail after the Earth's rotation and the gravity change are introduced.The refined pre-integration algorithm is applied to a multi-source fusion system based on tightly coupled GNSS/INS/vision.The experimental results show that the model error of the system pre-integration can be effectively reduced by using the refined pre-integration model,and the positioning and attitude accuracy of the multi-source fusion system is improved by 32.41%and 4.23%,respectively.

关 键 词：IMU预积分 地球自转 重力变化 图优化 多源融合定位 

分 类 号：V249.32[航空宇航科学与技术—飞行器设计] TP242[自动化与计算机技术—检测技术与自动化装置]