摘要
文中提出一种基于深度神经网络的焊缝跟踪系统机器人视觉标定方法,实现了机器人的简单快速标定。将焊缝跟踪传感器安装在机械臂末端,使用线激光器对被检测角点进行定位,工业相机拍摄对应的标定棋盘图像,用角点提取算法获取到相应棋盘格点的数字图像坐标,并通过示教器读取机械臂的各个关节角,利用神经网络极强的非线性映射能力,将其传入训练好的BP神经网络进行三维空间坐标的预测。此方法能够实现机器人的快速标定,避开传统标定方法中复杂的非线性运算,并减少坐标转换间的累积误差。实验结果表明,基于神经网络的标定方法具有较高的精度,且标定过程简单,为机器人视觉标定提供了一种新的方法。
A robot vision calibration method of weld seam tracking system based on depth neural network is proposed,which can realize the simple and fast calibration of the robot.The weld seam tracking sensor is installed at the end of the manipulator.The detected corner is positioned by the line laser.The industrial camera is used to obtain the corresponding checkerboard image.The digital pixel coordinates of the corresponding checkerboard grid point are got by the corner extraction algorithm.Each joint angle of the manipulator is read by the teaching pendant,and the strong nonlinear mapping ability of neural network is utilized to introduce it into the trained BP neural network to predict the three⁃dimensional spatial coordinates.This method can realize the rapid calibration of the robot,avoid the complex nonlinear operation in the traditional calibration method,and reduce the cumulative error during coordinate transformations.The experimental results show that the method based on neural network has high calibration accuracy and simple calibration process,which provides a new method for robot vision calibration.
作者
王嘉盛
张斌
应征
林子祥
WANG Jiasheng;ZHANG Bin;YING Zheng;LIN Zixiang(College of Metrology and Measurement Engineering,China Jiliang University,Hangzhou 310018,China;Zhejiang Academy of Special Equipment Inspection,Hangzhou 310009,China)
出处
《现代电子技术》
2023年第11期55-59,共5页
Modern Electronics Technique
关键词
视觉标定
焊缝跟踪系统
神经网络
机器人标定
像素点提取
空间坐标预测
vision calibration
weld seam tracking system
neural network
robot calibration
pixel point extraction
space coordinate prediction
