装配生产线的视觉AGV跟踪检测算法研究
|
摘要
为实现AGV对产线高精度实时跟踪,提出了一种基于机器视觉的跟踪检测方法。首先设计了一种红色圆形前景蓝色矩形背景的标识物挂载于装配生产线上,通过RGB-D相机对图像进行深度搜索及HSV空间上的滤波初步检测标识物是否存在于图像中,将标识物所在区域选作图像ROI区域;其次,对二值化后的图像使用边界跟踪算法寻找出圆形边界并剔除孤立点,使用RANSAC算法拟合出圆心;最后,结合生产线控制信号与坐标转换模型,使用信息滤波对标识物中心坐标进行更新。实验结果表明,该算法速度可达到20～25帧/秒,检测精度达到±1.5mm。
Online installation of parts on a moving assembly line can increase productivity. Tracking the production line with the follow-on AGV with the mounted parts not only avoids the modification of the site, but also meets flexible production. In order to realize the high-precision real-time tracking of AGV on the production line, a tracking method based on machine vision was proposed. Firstly, a marker with a red circular foreground blue rectangular background was designed to be mounted on the assembly line. The depth search was performed by the RGB-D camera and the filtering on the HSV space to detect whether the marker exists in the image and the marker is located. Then the located area of the marker was selected as the image ROI. Secondly, the boundary tracking algorithm was used to find the circular boundary and eliminated the outliers for the binary image, and the center of the circle was fitted by the RANSAC algorithm. Finally, information filtering was used to update the center coordinate of the marker in combination with the control signal and coordinate transformation model of the production line. The experimental results show that the speed of the algorithm can reach 20 ～ 25 frames per second with a detection accuracy of ±1.5 mm.
Online installation of parts on a moving assembly line can increase productivity. Tracking the production line with the follow-on AGV with the mounted parts not only avoids the modification of the site, but also meets flexible production. In order to realize the high-precision real-time tracking of AGV on the production line, a tracking method based on machine vision was proposed. Firstly, a marker with a red circular foreground blue rectangular background was designed to be mounted on the assembly line. The depth search was performed by the RGB-D camera and the filtering on the HSV space to detect whether the marker exists in the image and the marker is located. Then the located area of the marker was selected as the image ROI. Secondly, the boundary tracking algorithm was used to find the circular boundary and eliminated the outliers for the binary image, and the center of the circle was fitted by the RANSAC algorithm. Finally, information filtering was used to update the center coordinate of the marker in combination with the control signal and coordinate transformation model of the production line. The experimental results show that the speed of the algorithm can reach 20 ～ 25 frames per second with a detection accuracy of ±1.5 mm.
引文
[1] 武启平,金亚萍,任平,等.自动导引车(AGV)关键技术现状及其发展趋势[J].制造业自动化,2013,35(10):106-109,121.
[2] 张成龙,李少波,陈伟兴.面向离散制造物联车间的UWB实时定位系统设计[J].组合机床与自动化加工技术,2017(6):86-89.
[3] 蔡磊,周亭亭,郭云鹏,等.基于超声波定位的智能跟随小车[J].电子测量技术,2013,36(11):76-79,105.
[4] 徐文斌,蔡容华,林科荣,等.基于工控机的AGV装配机器人控制系统设计[J].电子技术应用,2013,39(7):131-133,136.
[5] 赵晨宇,陈息坤.差速转向农业专用AGV小车的设计与模糊控制研究[J].农机化研究,2016,38(11):123-127.
[6] 罗哉,唐颖奇,李冬,等.基于最优偏差路径的自动导引车纠偏方法[J].仪器仪表学报,2017,38(4):853-860.
[7] 石陈陈,楼佩煌,武星,等.自动导引车多摄像机主动导引系统的协同标定[J].仪器仪表学报,2014,35(11):2589-2599.
[8] 段振云,单紫薇,赵文辉,等.数字图像边缘的快速提取算法研究[J].组合机床与自动化加工技术,2017(12):12-14,20.
[9] 陈甦欣,魏前龙,王正阳,等.基于视觉的运动目标跟踪抓取和装配研究[J].机床与液压,2018,46(3):12-15.
[10] Amruta V Gulalkaril,Dongbo Shengl,Pandu Sandi Pratamal.Kinect Camera Sensor-based Object Tracking and Following of Four wheel Independent Steering Automatic Guided Vehicle Using Kalman Filter[A].In Proc.2015 15th International Conference on Control,Automation and Systems (ICCAS 2015)[C].Busan,IEEE,2015:1650-1655.
[2] 张成龙,李少波,陈伟兴.面向离散制造物联车间的UWB实时定位系统设计[J].组合机床与自动化加工技术,2017(6):86-89.
[3] 蔡磊,周亭亭,郭云鹏,等.基于超声波定位的智能跟随小车[J].电子测量技术,2013,36(11):76-79,105.
[4] 徐文斌,蔡容华,林科荣,等.基于工控机的AGV装配机器人控制系统设计[J].电子技术应用,2013,39(7):131-133,136.
[5] 赵晨宇,陈息坤.差速转向农业专用AGV小车的设计与模糊控制研究[J].农机化研究,2016,38(11):123-127.
[6] 罗哉,唐颖奇,李冬,等.基于最优偏差路径的自动导引车纠偏方法[J].仪器仪表学报,2017,38(4):853-860.
[7] 石陈陈,楼佩煌,武星,等.自动导引车多摄像机主动导引系统的协同标定[J].仪器仪表学报,2014,35(11):2589-2599.
[8] 段振云,单紫薇,赵文辉,等.数字图像边缘的快速提取算法研究[J].组合机床与自动化加工技术,2017(12):12-14,20.
[9] 陈甦欣,魏前龙,王正阳,等.基于视觉的运动目标跟踪抓取和装配研究[J].机床与液压,2018,46(3):12-15.
[10] Amruta V Gulalkaril,Dongbo Shengl,Pandu Sandi Pratamal.Kinect Camera Sensor-based Object Tracking and Following of Four wheel Independent Steering Automatic Guided Vehicle Using Kalman Filter[A].In Proc.2015 15th International Conference on Control,Automation and Systems (ICCAS 2015)[C].Busan,IEEE,2015:1650-1655.