Scholarlink

<p>To address the lack of flexibility in traditional welding robots when handling non-standard steel structures, a cascaded optimization algorithm framework is proposed. First, in the front-end recognition stage, an improved random sampling consensus algorithm driven by matching quality is constructed for accurate weld seam identification. In the back-end path planning stage, a fast exploratory random tree star algorithm is improved by fusing bidirectional growth of the exploratory tree with a greedy strategy to achieve efficient collision-free path planning. The results demonstrated that the improved random sampling consistency algorithm had a recognition error of 0.8 ± 0.1&#xa0;mm and a running time of only 68 ± 5 ms in low-noise scenes. The accuracy of inlier recognition remained stable at over 85%, and the reprojection error was as low as 2.3 pixels. In path planning simulation, the improved fast exploration random tree star algorithm had a much smaller exploration tree size than the comparison algorithms, generated the smoothest path, and had the best search efficiency and planning quality. In summary, the improved algorithm has shown significant advantages in both the accuracy of weld seam recognition and the efficiency of path planning. It provides key technological support for solving non-standard welding tasks, from environmental perception to autonomous operation.</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

RANSAC and Improved RRT* Algorithm-Driven Path Planning for Welding Robots and Intelligent Welding Process Optimization for Steel Structures

  • 期刊论文

  • 发表日期：

    2026年8月4日

查看全文
  • Shuhong Zhang,
  • Cuifeng Ren,
  • Fan Xu,
  • Yonghong Li,
  • Wenhao Li,
  • Xianyun Xu

摘要

To address the lack of flexibility in traditional welding robots when handling non-standard steel structures, a cascaded optimization algorithm framework is proposed. First, in the front-end recognition stage, an improved random sampling consensus algorithm driven by matching quality is constructed for accurate weld seam identification. In the back-end path planning stage, a fast exploratory random tree star algorithm is improved by fusing bidirectional growth of the exploratory tree with a greedy strategy to achieve efficient collision-free path planning. The results demonstrated that the improved random sampling consistency algorithm had a recognition error of 0.8 ± 0.1 mm and a running time of only 68 ± 5 ms in low-noise scenes. The accuracy of inlier recognition remained stable at over 85%, and the reprojection error was as low as 2.3 pixels. In path planning simulation, the improved fast exploration random tree star algorithm had a much smaller exploration tree size than the comparison algorithms, generated the smoothest path, and had the best search efficiency and planning quality. In summary, the improved algorithm has shown significant advantages in both the accuracy of weld seam recognition and the efficiency of path planning. It provides key technological support for solving non-standard welding tasks, from environmental perception to autonomous operation.