摘要
为了提高视觉导引自动导引车(automated guided vehicle,AGV)的定位精度,提出了一种多窗口实时测距的精确定位方法。该方法通过在定位点铺设圆形标识作为定位标识符,首先根据车载视觉系统结构,建立视觉实时测距模型。通过对获得的椭圆图像的几何特征进行提取,并采用基于曲率角估计方法对椭圆边缘精确识别;然后,基于最小二乘拟合直线获取椭圆中心位置。最后,结合采用多窗口图像处理方法,将定位过程分为多个阶段进行,逐步逼近,最终实现精确定位。实验结果表明,该方法能使得视觉导引AGV的定位精度达到2 mm。
In order to improve the positioning accuracy of visual navigation for Automated Guided Vehicle (AGV) , one type of precise positioning technology based on muki-windows and real-time distance measure is proposed. The approach realizes the positioning identifier by rounding the location point. Firstly, a visual real-time distance measure model is formulated by using tile on-board visual system. By combining the geometric feature of ellipse image and the curvature angle estimation, the ellipse margin is identified precisely. Secondly, the least squares is applied to fit a straight line for obtaining the ellipse center. Finally, the accurate positioning is achieved by utilizing muhi-window image processing which divides the positioning process into multi-steps and adopts step by step approximation. The experimental results show that the proposed approach can improve the AGV positioning accuracy to 2 mm.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2016年第6期1356-1363,共8页
Chinese Journal of Scientific Instrument
基金
江苏省科技支撑计划项目(BE2014137)
中国博士后科学基金资助项目(2015M580421)
汽车装配智能输送装备关键技术研发(BY2015003-11)
南京航空航天大学研究生创新基地(实验室)开放基金项目(KFJJ20150519)项目资助
关键词
自动导引车
视觉导引
实时测距
精确识别
automated guided vehicle ( AGV )
visual navigation
real-time distance measure
precise recognition