Scholarlink

<p>With the rapid development of intelligent welding, the deep integration of AI and machine vision to achieve initial weld point (IWP) and weld positioning, as well as weld path planning have become an important research direction in recent years. This paper proposed a method based on passive light and active light vision to realize IWP guidance and weld path planning. The camera is guided to two specific locations to capture the complete workpiece. In the passive light environment, the improved Yolov8 is utilized to recognize the initial weld region (IWR), and IWPs are directly recognized with the help of the Centre-box and substituted into “monocular vision dual-position shooting” stereoscopic vision model to achieve approximate IWP positioning. The camera is then guided to the relative position of approximate IWP by structured light planes, image processing is utilized to extract feature points and rectify the camera posture. Weld depth with higher precision is extracted with improved camera posture. Yolov8 is utilized again to recognize and precisely position the IWP. Based on the weld features, the welding torch posture is planned. Experimental results demonstrate that the method can precisely recognize various types of weld seams, with the IWP and weld seam positioning accuracy within 0.9&#xa0;mm, meeting the precision requirements for subsequent welding operations.</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

A Novel Initial Weld Point Positioning and Path Planning Method for Robot Welding Based on Passive light stereo vision and Structured Light Vision

  • 期刊论文

  • 发表日期：

    2025年6月21日

查看全文
  • Tianqi Wang,
  • Qiaochu Li,
  • Junjie He,
  • Xiao Li

摘要

With the rapid development of intelligent welding, the deep integration of AI and machine vision to achieve initial weld point (IWP) and weld positioning, as well as weld path planning have become an important research direction in recent years. This paper proposed a method based on passive light and active light vision to realize IWP guidance and weld path planning. The camera is guided to two specific locations to capture the complete workpiece. In the passive light environment, the improved Yolov8 is utilized to recognize the initial weld region (IWR), and IWPs are directly recognized with the help of the Centre-box and substituted into “monocular vision dual-position shooting” stereoscopic vision model to achieve approximate IWP positioning. The camera is then guided to the relative position of approximate IWP by structured light planes, image processing is utilized to extract feature points and rectify the camera posture. Weld depth with higher precision is extracted with improved camera posture. Yolov8 is utilized again to recognize and precisely position the IWP. Based on the weld features, the welding torch posture is planned. Experimental results demonstrate that the method can precisely recognize various types of weld seams, with the IWP and weld seam positioning accuracy within 0.9 mm, meeting the precision requirements for subsequent welding operations.