基于图像识别的机器人定位与导航算法的研究  

The Research on Robot Positioning and Navigation Algorithm Based on Image Recognition

作　　者：戴月[1] DAI Yue(Chuzhou Vocational And Technical College,Chuzhou Anhui 239000,China)

机构地区：[1]滁州职业技术学院,安徽滁州239000

出　　处：《长春工程学院学报（自然科学版）》2020年第1期105-109,共5页Journal of Changchun Institute of Technology：Natural Sciences Edition

摘　　要：近年来,由于移动机器人逐渐脱离实验环境进入到商用阶段,采用激光雷达进行地图导航的方式由于成本过高逐渐被低成本的视觉方案所取代。依托两台单目摄像机,基于图像识别理论可以对机器人的工作环境进行感知建模,所获得的信息相较于传统的激光雷达的距离信息更增加了特征信息,使得机器人的导航定位更加灵活。针对移动机器人的导航问题,提出了一种基于图像识别的定位导航算法。首先,分析了问题的研究背景和研究现状;其次,构建了双目视觉定位系统并分析了图像特征提取的主要方法;再次,提出了一种改进的图像匹配方法实现视觉SLAM;最终,所设计的算法通过实验进行了验证。实验结果表明,所设计的算法能够准确呈现工作环境的原始信息。In recent years,as mobile robots have gradually moved away from the experimental environment into the commercial phase,the way of using Lidar for map navigation with high cost has been replaced by low-cost visual solutions.Based on two monocular cameras,the image recognition theory can be used to model the working environment of the robots.The obtained information adds more characteristic information than the distance information of the traditional laser radar,which makes the robot’s navigation positioning more flexible.Aiming at the navigation problem of mobile robots,this paper proposes a positioning and navigation algorithm based on image recognition.Firstly,the research background and research status of the problem are analyzed.Secondly,the binocular vision localization system is constructed,and the main methods of image feature extraction are analyzed.Thirdly,an improved image matching method is proposed to realize visual SLAM.Finally,the designed algorithm is verified by experiments.The experimental results show that the designed algorithm can accurately present the original information of the working environment,and has certain practical value.

关 键 词：移动机器人 视觉 导航 定位 

分 类 号：TP242.3[自动化与计算机技术—检测技术与自动化装置]