An adaptive welding method and system based on visual perception

By employing visual perception and adaptive welding methods, and utilizing 3D structured light vision sensors and algorithms to fit weld seam feature surfaces, the programming difficulties of welding robots in non-standard production and complex weld seam scenarios have been solved, achieving efficient and stable welding results.

CN122199588APending Publication Date: 2026-06-12CHINA SHIPPING IND JIANGSU
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
CHINA SHIPPING IND JIANGSU
Filing Date
2026-03-13
Publication Date
2026-06-12

AI Technical Summary

⚠Technical Problem

Existing welding robots struggle to adapt to changes in workpieces in non-standard production and complex weld seam scenarios, requiring frequent reprogramming, resulting in high costs and unstable welding quality, especially inefficient in small-batch, multi-variety production.

⚗Method used

An adaptive welding method based on vision perception is adopted. The weld seam is scanned by a 3D structured light vision sensor, and the weld seam feature surface is fitted by the least squares method and KD_TREE algorithm to generate a high-precision welding trajectory in real time. The welding parameters are optimized by using a process database.

🎯Benefits of technology

It enables rapid and accurate welding path planning, reduces programming time and manual labor dependence, improves welding quality and equipment utilization, and adapts to the needs of small-batch, multi-variety production.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122199588A_ABST
    Figure CN122199588A_ABST
Patent Text Reader

Abstract

The application discloses a kind of self-adapting welding method and system based on visual perception, and the end of welding robot is configured 3D structured light vision sensor to scan target welding area and determine three-dimensional accessible domain range and obtain the weld three-dimensional point cloud data of weld execution area, by least square method fitting into multiple independent planes and calculating normal vector, based on normal vector identification as feature surface or non-feature surface and calculate groove angle and plate thickness, by KD_TREE find the nearest point pair and apply least square method fitting into preliminary center line, project to theoretical intersection line and obtain weld center line.After correction and optimization based on groove angle, plate thickness and assembly gap, combined with process database matching welding process requirement to generate welding trajectory and issue execution welding.Visual perception replaces artificial, and the integrated 3D vision system can quickly scan workpiece, automatically identify weld position, groove shape and assembly gap, and real-time generate high-precision three-dimensional welding trajectory.
Need to check novelty before this filing date? Find Prior Art