Scholarlink

<p>As a critical component of the turbine, blade manufacturing quality directly impacts energy conversion efficiency. During the trajectory planning phase of robotic blade grinding, improperly established machining coordinate systems can easily lead to systematic errors, resulting in uneven machining allowances and deviations from the design specifications. This paper applies point cloud registration to align key datum surfaces and objects from the 3D laser scan, establishing an accurate coordinate system for the robot end fixture under actual clamping condition. The proposed improved RANSAC algorithm effectively leverages the location information of all inlier points while minimizing noise interference, ensuring the accuracy and stability in feature extraction. Additionally, the stepwise constraint strategy for obtaining the initial position of fine registration significantly enhances the adaptability of the ICP registration algorithm, improving final registration quality. Point cloud data processed using software developed with the Point Cloud Library (PCL) validates the method’s effectiveness and stability. Experiments demonstrate that the method in this paper greatly improves the efficiency and quality of mass and multi-model blade grinding, offering valuable insights for other complex surface grinding applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Research on the calibration method of robot grinding system with stepwise constraint of key datum based on point cloud technology

  • 期刊论文

  • 发表日期：

    2025年5月8日

查看全文
  • You Zhou,
  • Jinhua Zhang,
  • Zhenhua Jiang,
  • Jun Hu

摘要

As a critical component of the turbine, blade manufacturing quality directly impacts energy conversion efficiency. During the trajectory planning phase of robotic blade grinding, improperly established machining coordinate systems can easily lead to systematic errors, resulting in uneven machining allowances and deviations from the design specifications. This paper applies point cloud registration to align key datum surfaces and objects from the 3D laser scan, establishing an accurate coordinate system for the robot end fixture under actual clamping condition. The proposed improved RANSAC algorithm effectively leverages the location information of all inlier points while minimizing noise interference, ensuring the accuracy and stability in feature extraction. Additionally, the stepwise constraint strategy for obtaining the initial position of fine registration significantly enhances the adaptability of the ICP registration algorithm, improving final registration quality. Point cloud data processed using software developed with the Point Cloud Library (PCL) validates the method’s effectiveness and stability. Experiments demonstrate that the method in this paper greatly improves the efficiency and quality of mass and multi-model blade grinding, offering valuable insights for other complex surface grinding applications.