Weld seam recognition algorithm based on a fast point cloud plane fitting method.

Area of Science:

• Robotics and Automation
• Computer Vision
• Materials Science

Background:

• Existing point cloud weld extraction algorithms for 3D vision-based robotic intelligent welding suffer from low accuracy and slow processing speeds.
• Efficient and accurate weld seam extraction is crucial for automated manufacturing processes.

Purpose of the Study:

• To propose a novel, three-stage automatic point cloud weld extraction method to address the limitations of existing algorithms.
• To enhance the accuracy, speed, and robustness of weld seam extraction in 3D vision-guided robotic welding.

Main Methods:

• Improved Random Sample Consensus (RANSAC) for enhanced plane segmentation via local neighborhood sampling and dynamic curvature detection.
• Weld seam feature point extraction using plane intersection and distance threshold methods.
• Farthest Point Sampling (FPS) for denoising and resampling, followed by curve fitting for high-precision contour reconstruction.

Main Results:

• The proposed method significantly improves the efficiency of plane fitting.
• Accurate weld seam feature points are obtained using the developed extraction techniques.
• High-precision weld contour reconstruction is achieved through denoising, resampling, and curve fitting.

Conclusions:

• The novel three-stage method demonstrates high efficiency, robustness, and engineering adaptability for point cloud weld extraction.
• This approach offers a promising solution for advancing 3D vision-based robotic intelligent welding systems.