成捆圆钢端面点云处理与标牌焊接点定位方法研究  

Research on point cloud processing of end surface of bundled round steels and positioning method of welding tag point

作　　者：李文忠[1] 王者涵 张付祥[1] 王春梅 黄风山[1] LI Wenzhong;WANG Zhehan;ZHANG Fuxiang;WANG Chunmei;HUANG Fengshan(School of Mechanical Engineering,Hebei University of Science and Technology,Shijiazhuang,Hebei 050018,China;School of Electrical Engineering,Hebei University of Science and Technology,Shijiazhuang,Hebei 050018,China)

机构地区：[1]河北科技大学机械工程学院,河北石家庄050018 [2]河北科技大学电气工程学院,河北石家庄050018

出　　处：《河北工业科技》2024年第6期470-480,共11页Hebei Journal of Industrial Science and Technology

基　　金：中央引导地方科技发展资金项目(246Z1808G);石家庄市驻冀高校重大科技专项(241080507A)。

摘　　要：为了实现标牌焊接机器人作业准确定位,提出了一种成捆圆钢端面点云处理及焊接点定位方法。首先,在搜索规则、参数调整等方面对传统点云数据滤波方法进行了改进;其次,提出了融合欧式聚类、AABB(axis-aligned bounding box)算法和RKNN(reverse K-nearest neighbor)搜索算法的点云分割算法;再次,考虑焊牌要求制定了焊接点位选取策略,通过法向量估计进行点云姿态校正,采用RANSAC算法对选定圆钢端面点云进行拟合,得到焊接中心点位;最后,进行了焊牌实验。结果表明:该方法有效去除了无关点云的影响,缩短了滤波时间,大大减少了点云数量,有效地对粘连点云进行分割,得到了每根圆钢端面点云,确定了满足焊牌条件的圆钢,并准确得到了焊接点的坐标;焊牌实验表明,相对定位误差小于8%,满足实际生产需求。所提方法有利于提高焊牌机器人系统的焊牌定位准确度和工作效率,可为类似工业机器人系统定位提供参考。In order to achieve accurate positioning of welding tag by robot system,a processing method of point cloud of the end surface of bundled round steels and welding point positioning was proposed.Firstly,the traditional filtering methods of point cloud were improved in terms of search rules and parameter.Secondly,combining Euclidean clustering,axis-aligned bounding box(AABB)and reverse K-nearest neighbor(RKNN)search,a point cloud segmentation algorithm was proposed.Then,a welding point selection strategy was formulated based on welding tag requirements.The point cloud position was corrected through normal vectors estimation.The random sample consensus(RANSAC)algorithm was used to fit the seleted point cloud of the end surface of bundled round steels to obtain the welding point position.Finally,a welding tag experiment was conducted.The results show that this method effectively removes the influence of irrelevant point clouds,shortens the filtering time,greatly reduces the number of point cloud,effectively segments the point cloud adhesion,obtains the point cloud for each round steel end face,determines the round steel that meets the welding tag requirements,and accurately obtains the coordinates of the welding points.The welding plate experiment shows that the relative positioning error is less than 8%,which meets the actual production needs.The proposed method is conducive to improve the accuracy and efficiency of welding tag positioning in the welding tag robot system,and can also provide reference for the positioning of similar industrial robot systems.

关 键 词：计算机图象处理 点云 改进滤波方法 RKNN算法 成捆圆钢 焊牌 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]