室内外混合环境下基于IMM-EKF的AGV连续定位方法研究  

Research on continuous positioning method of AGV based on IMM-EKF in indoor and outdoor mixed environment

作　　者：钱伟 陈析[2] 任雪林 孙丙宇 罗强[1] 王海宝 QIAN Wei;CHEN Xi;REN Xuelin;SUN Bingyu;LUO Qiang;WANG Haibao(School of Mechanical Engineering,Chongqing Three Gorges University,Chongqing 404130,China;Hefei Institute of Material Science,Chinese Academy of Sciences,Hefei 230031,China)

机构地区：[1]重庆三峡学院机械工程学院,重庆404130 [2]中国科学院合肥物质科学研究院,安徽合肥230031

出　　处：《传感器与微系统》2023年第7期61-65,共5页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(61773360);2019年重庆市人工智能+智慧农业学科群开放基金资助项目(ZNNYKFA201901)。

摘　　要：针对室内外混合环境下自动导引车(AGV)连续定位中存在多模型不匹配竞争、定位精度差的问题,提出一种基于交互式多模型—扩展卡尔曼滤波(IMM-EKF)的AGV室内外连续定位算法。针对AGV连续定位存在定位精度差的问题,提出采用平行扩展卡尔曼滤波器分别实现激光雷达(LiDAR)/里程计(ODOM)、全球导航卫星系统(GNSS)/ODOM的融合滤波定位。针对AGV连续定位存在多模型不匹配竞争问题,提出通过模型的似然概率分别计算LiDAR和GNSS模型概率,并根据模型概率对定位结果进行加权融合,从而计算AGV的最优位姿估计。AGV连续定位实验结果表明:本文提出融合LiDAR/GNSS/ODOM的IMM-EKF连续定位滤波算法,极大地提高了室内外连续定位精度、并有效抑制模型间的不匹配竞争关系,实现AGV的实时全局精准定位。Aiming at the problems of continuous positioning of automatic guided vehicle(AGV)in indoor and outdoor mixed environments,such as model mismatch competition and poor positioning precision,an AGV indoor and outdoor continuous positioning algorithm based on interactive multi-model-extended Kalman filtering(IMM-EKF)is proposed.Aiming at the problem of poor positioning precision continuous positioning,parallel extended Kalman filters are proposed to use to achieve LiDAR/ODOM and GNSS/ODOM fusion filtering positioning respectively.Aiming at the problem of model mismatch competition in continuous positioning of AGV,the likelihood probabilities of the two models are proposed to calculate the LiDAR and GNSS model probabilities respectively,and the positioning results are weighted fusion according to the model probabilities,so as to calculate the optimal pose estimation of the AGV.The results of AGV continuous positioning experiments show that the IMM-EKF continuous positioning filtering algorithm is proposed to fuse with LiDAR/GNSS/ODOM,which greatly improves the precision of indoor and outdoor continuous positioning,and effectively suppresses the mismatched competition relationship between models,and realizes the performance of AGV real-time global precise positioning.

关 键 词：自动导引车 室内外连续定位 交互式多模型 多传感器融合 扩展卡尔曼滤波 

分 类 号：TP393[自动化与计算机技术—计算机应用技术] TP212[自动化与计算机技术—计算机科学与技术]