摘要
为提高工业机器人在复杂化工生产环境下的目标识别与抓取精度,对化工生产线工业机器人视觉点位修正校准方案进行优化研究。提出一种基于3D双目视觉的立体校准修正算法,通过对极几何模型、KNN算法以及点云预处理,实现对化工生产线工业机器人的六自由度位姿估算。结果表明,改进的视觉校准和姿态估算方法能够有效提升机器人在复杂环境下的抓取精度,抓取误差在3 mm以内。所提出的视觉修正算法能够稳定输出高置信度的目标位姿信息,在实际的工业应用中具有可行性。
In order to improve the target recognition and grasping accuracy of industrial robots in complex chemical production environment,the visual point correction and calibration scheme of industrial robots in chemical production line was optimized.A stereo calibration and correction algorithm based on 3D binocular vision was proposed.The 6-DOF pose estimation of industrial robots in chemical production line was realized through polar geometry model,KNN algorithm and point cloud preprocessing.The results showed that the improved visual calibration and attitude estimation methods effectively improved the grasp accuracy of the robot in the complex environment,and the grasp error was less than 3 mm.The proposed visual correction algorithm can output target pose information with high confidence stably and is feasible in practical industrial applications.
作者
王琳
赵学强
WANG Lin;ZHAO Xueqiang(Yantai Vocational College,Yantai 264000,Shandong China;Shandong University of Technology,Zibo 255000,Shandong China)
出处
《粘接》
CAS
2024年第9期65-67,71,共4页
Adhesion
基金
山东省教育教学研究课题项目(项目编号:2021JXY125)。
