摘要
针对自动导航小车存在定位不精准的问题,通过机器视觉方法来校正小车位姿。从源图像中识别出定位孔的像素坐标并计算转换到全局坐标系下,通过多组匹配点计算小车实际位姿。在分析相机标定与位姿计算过程中可能引入误差的基础上,提出相应误差消除策略。通过实验,验证了视觉标定结果的可靠性和误差消除算法的正确性,为视觉引导下的机器人装配工作提供了依据。
The visual positioning method here is targeted to deal with the inaccurate positioning issue of AGV. The pixel coordinates of the positioning holes from the source image were identified, calculated and transferred into the coordinate system. The actual position of AGV was calculated through multiple sets of matching points. An error canceling algorithms was proposed based on the possible errors analyzed in the process of calibration and positioning calculation. The experiment result proves the visual calibration is reliable and the error canceling algorithms is correct, so it is useful to guide the robot’s visual assembling.
作者
吴广雨
张伟军
杨保佳
WU Guangyu;ZHANG Weijun;YANG Baojia(School of Mechanical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China)
出处
《机械与电子》
2020年第3期75-80,共6页
Machinery & Electronics
关键词
机器视觉
手眼标定
机器人
位置校正
柔性生产
machine vision
hand-eye calibration
robotics
position correction
flexible production
