面向AGV导航的双目立体相机研究  

Research on Binocular Stereo Camera for AGV Navigation

作　　者：魏军[1] 栾金圆 胡岩 穆庆晟 Wei jun;Luan Jinyuan;Hu Yan;Mu Qingsheng(School of Electrical Engineering,University of Jinan,Jinan 250022,China;Shandong Intelligent Robot Application Technology Research Institute,Jining,Shandong 273500,China)

机构地区：[1]济南大学自动化与电气工程学院,济南250022 [2]山东省智能机器人应用技术研究院,山东济宁273500

出　　处：《机电工程技术》2023年第6期14-18,共5页Mechanical & Electrical Engineering Technology

基　　金：山东省科技型中小企业创新能力提升工程项目(2021TSGC1384)。

摘　　要：根据使用场景及应用需求,提出采用X角点构成视觉标签,利用视觉标签标记场景中关键路径点的方式实现AGV的定位和路径规划的视觉导航方法。搭建双目立体相机,基于段测试X角点检测算法,识别视觉标签、建立标签坐标系统,根据视觉标签库中的预存信息检索、确定AGV的地图位置;基于视觉标签局部坐标系统,确定AGV的姿态信息,在简化导航复杂度的同时,保证AGV路径控制和运行控制精度和可靠性,进而实现基于视觉标签的自动导航和运动控制。将双目立体相机应用于AGV现场试验,结果表明,所搭建的双目立体相机能够为AGV的视觉导航提供信息,开发的基于视觉标签的导航算法具有较好的鲁棒性,视觉标签的正确识别率达95%,双目立体相机定位误差小于2%,充分发挥了视觉导航的优势,能够完成园区内的可靠自动运行。Based on application requirements and usage scenarios,a visual navigation method is proposed to use X corner to form visual labels,and to use visual labels to mark key path points in the scene to achieve AGV positioning and path planning.Binocular stereo camera is built,the X corner detection algorithm is tested based on segments,visual labels are then identified,and a label coordinate system is established,based on the pre-stored information in the visual label library,the map position of AGV can be retrieved and determined.Based on the visual label local coordinate system,the attitude information of AGV is determined,which simplifies the navigation complexity while ensuring the accuracy and reliability of AGV path control and operation control,thereby achieving automatic navigation and motion control based on visual labels.The binocular stereo camera is applied to AGV on-site for testing,and the results show that the constructed binocular stereo camera can provide information for AGV visual navigation.The developed navigation algorithm based on visual labels is fairly robust,the correct recognition rate of visual labels reaches 95%,and the positioning error of binocular stereo cameras is less than 2%,fully leveraging the advantages of visual navigation,and achieving reliable automatic operation in the park with visual labels.

关 键 词：立体相机 AGV 三维定位 图像处理 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]