收稿日期:2021-12-24修回日期:2022-01-03第40卷第2期计算机仿真2023年2月文章编号:1006-9348(2023)02-0466-05基于视觉双窗口的机器人视觉跟随式导航仿真王连枝,朱朝旭,赵帅,李宝玺(开封大学电子电气工程学院,河南开封475000)摘要:针对未知环境下机器人自主导航问题,以提升机器人视觉跟随精度与自主导航性能为目的,研究基于视觉双窗口的机器人视觉跟随式导航方法。机器人通过视觉双窗口摄像头采集跟随对象图像,采用两步标定算法获取视觉双窗口摄像头内外参数,将双窗口摄像头统一至相同世界坐标系内,实现视觉双窗口摄像机图像与现实场景内图像间的映射。对采集的图像序列进行特征立体匹配,将二维与三维扩展Kalman滤波器同时用于跟随对象位置预测,实现跟随目标的跟随定位。以基于行为的控制结构作为机器人导航控制结构的基础,根据跟随目标的跟随定位结果利用宏行为控制机器人运动。仿真结果显示,上述方法可控制机器人在复杂背景环境下准确跟随设定的对象,在存在障碍物的条件下,机器人可准确避开障碍物,并在之后回归到准确的跟随导航路线。关键词:视觉双窗口;机器人视觉;跟随式导航;摄像头标定;位置预测;宏行为中图分类号:TP391文献标识码:BSimulationofRobotVisualFollowingNavigationSimulationofRobotBasedonVisualDoubleWindowsWANGLian-zhi,ZHUZhao-xu,ZHAOShuai,LIBao-xi(DepartmentoftheElectricalEngineering,KaifengUniversity,KaifengHenan475000,China)ABSTRACT:Aimingattheproblemofrobotautonomousnavigationinunknownenvironment,inordertoimprovetherobotvisualfollowingaccuracyandautonomousnavigationperformance,arobotvisualfollowingnavigationmethodbasedonvisualdoublewindowsiswasstudied.Therobotcollectedsthefollowingobjectimagethroughthevisualdualwindowcamera,obtainsobtainedtheinternalandexternalparametersofthevisualdualwindowcamerabyusingthetwo-stepcalibrationalgorithm,unifiesunifiedthedualwindowcameraintothesameworldcoordinatesystem,andrealizesrealizedthemappingbetweenthevisualdualwindowcameraimageandtheimageintherealscene.Onthisbasis,thefeaturestereomatchingiswascarriedoutforthecollectedimagesequence,andthetwo-dimensionalandthree-dimensionalextendedKalmanfilterwereareusedtopredictthepositionofthefollowingobjectatthesame...