基于图优化的惯性/地磁/激光雷达复合定位技术研究

为解决卫星失锁条件下车辆的高精度、高可靠性定位问题，结合惯性导航、运动约束、地磁匹配、激光点云匹配方法的优势，提出了一种基于图优化的惯性/地磁/激光雷达复合定位方案。首先，采用因子图优化算法，实现惯导预积分信息、运动约束信息、地磁匹配信息、激光雷达点云匹配信息的异步多源信息融合。然后，在点云匹配失效时，惯性/运动约束/地磁可以实现精度保持，为点云匹配提供持续可用的先验信息，以避免点云匹配一旦失效后由于先验位姿发散再难匹配的问题。最后，搭建试验平台完成跑车试验。试验结果表明，该方案可以实现车辆的高精度、高可靠性定位，点云匹配有效定位精度为1.24m(max),均方根为0.48m,在因点云地图缺失而造成匹配失效时，惯性/运动约束/地磁可实现定位精度保持在10.29m(max)。

Research on Inertial/Geomagnetic/LiDAR Integrated Positioning Technology Based on Graph Optimization

Inordertosolvetheproblemofhigh-precisionpositioningofvehiclesundertheconditionof satellitelock-out,basedonthecomplementaryadvantagesofinertialnavigation,motionconstraints,geomagneticmatchingandLiDARpointcloudmatchingmethods, anovelinertial/geomagnetic/LiDARintegratedpositioningschemeusinggraphoptimizationisproposed. Firstly,thefactorgraphoptimizationalgorithmisusedtorealizetheintegrationofasynchronousmulti- sourceinformation,includinginertialpreintegrationinformation, motionconstraintinformation,geomagneticmatchinginformationandLiDAR pointcloudmatchinginformation.Then,inthecaseofinvalidpointcloudmatching,inertial/motionconstraint/ geomagneticnavigationcanachieveaccuracymaintenance,soastoprovidecontinuouslyavailable priorinformationforpointcloudmatching,whichcanavoidtherematchproblemforLiDARinthiscase. Finally,atestplatformisbuilttocompletevehicleroadtests.Theresultsshowthattheproposedscheme canachievehighprecisionandhighreliabilitynavigation.Thepositioningaccuracyofpointcloudmatching methodis1.24m(max),theroot-mean-squareis0.48m,andthepositioningaccuracyofinertial/motion constraint/geomagneticnavigationmethodis10.29m(max)whenthepointcloudmatchingfailsforshort ofpointcloudmap.