YOLO-LaserGalvo: A Vision-Laser-Ranging System for High-Precision Welding Torch Localization

Area of Science:

• Robotics and Automation
• Computer Vision
• Sensor Fusion

Background:

• Precise localization of welding torch tips is critical for industrial automation.
• Existing methods often struggle with complex illumination and require artificial markers.

Purpose of the Study:

• To introduce a novel closed-loop visual positioning system (YOLO-LaserGalvo) for accurate welding torch tip localization.
• To evaluate its performance against traditional methods under challenging conditions.

Main Methods:

• Integration of a monocular camera, infrared laser distance sensor with galvanometer scanner, and a custom YOLOv11 deep learning model.
• Real-time 3D position computation by fusing vision-based detection with laser-measured distance.
• Marker-free approach eliminating the need for structured lighting or artificial markers.

Main Results:

• The YOLO-LaserGalvo system demonstrated superior robustness and accuracy compared to color-marker and ArUco baselines.
• Achieved high-precision torch positioning in complex illumination scenarios.
• Outperformed traditional methods in accuracy, robustness, and processing speed.

Conclusions:

• The proposed YOLO-LaserGalvo system offers a marker-free, high-precision solution for welding torch tip localization in industrial automation.
• It provides significant advantages in accuracy, robustness, and speed over existing techniques.
• Potential for future enhancements includes 6-DOF pose estimation and multi-sensor integration.