摘要
为解决数控机床刀具无拆卸条件下,基于机械臂的刀具损伤视觉检测系统对准对焦调节耗时长、计算分析方法鲁棒性差等难题,提出了一种融合YOLOv5网络智能感兴趣区域(region of interest, ROI)的机器人视觉系统自动对准对焦方法。首先,利用ROI模型检测并定位刀具中心,通过九点标定法计算机械臂末端对准坐标;然后自适应筛选ROI对焦窗口,采用改进的Laplacian函数计算清晰度评价值以确定最佳刀具图像。在实际设备上开展实验后结果表明,所提方法比一般方法灵敏度至少提高1.63倍,平均中心点误差为3.76像素,有效提升了刀具损伤视觉检测系统的准确度和灵活性。
Under the condition of no disassembly of CNC machine tools,to address the challenges of time-consuming alignment and focusing adjustment in the tool damage visual detection system based on robotic arm,along with poor robustness of the calculation analysis method,an automatic alignment and focusing method for a robot vision system that integrates the YOLOv5 network for intelligent region of interest(ROI)fusion is proposed.Firstly,the ROI model is utilized to detect and locate the center of the tool,and the coordinates for the end effector of the robotic arm are calculated using the nine-point calibration method.Subsequently,an adaptive selection of the ROI focusing window is performed,and an improved Laplacian function is employed to compute the sharpness evaluation value for determining the best tool image.Experimental results conducted on actual equipment demonstrate that the proposed method enhances sensitivity by at least 1.63 times compared to conventional methods,with an average center point error of 3.76 pixels,effectively improving the accuracy and flexibility of the tool damage visual detection system.
作者
郝玲
潘一
何建樑
孟林坤
魏宇康
王禹林
HAO Ling;PAN Yi;HE Jianliang;MENG Linkun;WEI Yukang;WANG Yulin(School of Mechanical Engineering,Nanjing University of Science and Technology,Nanjing 210094,CHN)
出处
《制造技术与机床》
北大核心
2024年第8期72-77,共6页
Manufacturing Technology & Machine Tool
基金
国家自然科学基金资助项目(52075267)。
关键词
刀具损伤
视觉检测
自动对焦
自动对准
机械臂
tool damage
visual inspection
autofocus
automatic alignment
robotic arm
作者简介
第一作者:郝玲,女,2000年生,硕士研究生,研究方向为刀具失效检测。E-mail:hl0416@njust.edu.cn;通信作者:王禹林,男,1981年生,博士,教授,博士研究生导师,研究方向为先进制造技术、精密测控技术、智能装备与机器人。E-mail:wyl_sjtu@126.com。
