电子测量技术ELECTRONICMEASUREMENTTECHNOLOGY第45卷第23期2022年12月DOI:10.19651/j.cnki.emt.2209998基于图优化和EKF的UWB组合定位算法*易颂轩1高伟强1刘建群1郭俊权1刘达2(1.广东工业大学机电工程学院广州510006;2.佛山科莱机器人有限公司佛山528225)摘要:针对超宽带(UWB)技术受电气转换时间延迟,天线延迟以及基站间时间同步等因素的影响而导致UWB传感器的测量值存在误差的问题,本文提出一种基于图优化和EKF的UWB组合定位算法,首先使用EKF算法得到定位初值,接着构建包含UWB测距误差的目标函数,然后通过图优化算法求解使整个定位过程UWB总体测距误差最小的基站位置,然后将求解的基站位置代入EKF算法进行第二次计算,得到更为精确的定位结果,最后应用Mean-Shift算法对定位结果进行聚类分析。实验结果表明:本文提出算法相对于EKF算法和最小二乘法,水平定位精度平均提高了27%和38%,而且得到的目标运动轨迹平滑性远优于传统定位算法。关键词:超宽带;室内定位;图优化;扩展卡尔曼滤波;Mean-Shift聚类中图分类号:TP212.9文献标识码:A国家标准学科分类代码:510.1050UWBcombinedlocationalgorithmbasedongraphoptimizationandEKFYiSongxuan1GaoWeiqiang1LiuJianqun1GuoJunquan1LiuDa2(1.SchoolofElectromechanicalEngineering,GuangdongUniversityofTechnology,Guangzhou510006,China;2.FoshanClevRobotCo.,Ltd.,Foshan528225,China)Abstract:AimingattheproblemthatUWBsensormeasurementerrorsarecausedbyelectricalconversiontimedelayandantennadelayinUWBtechnology,aUWBcombinedpositioningalgorithmbasedongraphoptimizationandEKFisproposedinthispaper.Firstly,EKFalgorithmisusedtoobtaintheinitialpositioningvalue,andthentheobjectivefunctionincludingUWBrangingerrorisconstructed.Then,thegraphoptimizationalgorithmwasusedtosolvethebasestationlocationthatminimizedtheoveralllocationerrorofUWBinthewholepositioningprocess.Then,thesolvedbasestationlocationwassubstitutedintoEKFalgorithmforthesecondcalculationtoobtainmoreaccuratelocationresults.Finally,Mean-Shiftalgorithmwasusedtoperformclusteranalysisonthelocationresults.ExperimentalresultsshowthatcomparedwiththeEKFalgorithmandtheleastsquaremethod,thehorizontalpositioningaccuracyoftheproposedalgorithmisimprovedby27%and38%onaverage,andtheobtainedtar...
