Machine Vision Control of Laser Weld Processing
Robotics & Automation Engineering

Providing clients with exemplary service that often leads to repeat business and long-term relationships is one goal of Southwest Research Institute (SwRI). An industrial manufacturer chose to approach SwRI for assistance with a challenging machine vision problem, and the decision to work with us was based on the client relationship built more than a decade earlier.

Laser Welding Process Background

The client uses a number of tube-forming machines to fabricate long lengths of steel tubing from raw strip material stock. A series of rollers shape the strip into the appropriate tube shape, and the seam is closed using a laser welding system. To ensure consistent quality of the final product, the laser must be precisely aligned on the tube seam.

However, due to material variations, the tube seam position drifts over time relative to the laser welding point. This misalignment leads to decreased tube strength and in the most extreme cases, even gaps in the welded seam. The original tube-forming station contained an actuator and camera system to allow the operator to manually monitor and control the weld alignment, but it was insufficient to meet increasing quality targets demanded of the process.

A camera and lighting system are used to image the weld position for machine-vision control of a laser-welding process.

image of A camera and lighting system are used to image the weld position for machine-vision control of a laser-welding process.

A camera and lighting system are used to image the weld position for machine-vision control of a laser-welding process.

Machine Vision as a Solution

SwRI developed a machine-vision solution to provide real-time monitoring and control of the laser/seam alignment. The existing camera, lighting, and actuator components were updated with drop-in replacements suitable to a machine-vision system. A PC-based Operator Interface Terminal (OIT) provided both process display and machine-vision control functions. The system used the machine-vision results to provide overlay enhancements on the process display and statistical process data for offline analysis.

This problem was a particularly challenging machine-vision application due to the high contrast difference between the weld spot and tube seam features. Smoke, flare-ups, and other process-related anomalies provided additional complications to machine-vision analysis. Typically, process alignment applications can often be solved by generic off-the-shelf vision solutions. The unique tracking challenges of this problem proved a good match for our real-world machine-vision expertise when an off-the-shelf solution was not feasible.

Results/Outcome of Vision-Based Laser Alignment System

This system was demonstrated to provide substantial improvements in both the quality and consistency of the tube-welding process: Tube strength was improved due to better real-time laser/seam alignment. Weld gaps were nearly eliminated due to the system’s long-term drift compensation and quick tracking of short-term alignment changes. As an unexpected benefit, the system’s weld-alignment statistics functionality has proven to be one of the most valuable analytical tools for process management. Based on the success of the initial prototype system, the client has fabricated and installed replicate systems on each of their remaining tube-forming machines. Our typical product development process provides all necessary documentation, drawings and software to the client along with delivery of the prototype system.

Related Terminology

machine vision  •  real-time monitoring  •  vision-based laser alignment  •  product development  •  prototype  •  laser welding

Benefiting government, industry and the public through innovative science and technology
Southwest Research Institute® (SwRI®), headquartered in San Antonio, Texas, is a multidisciplinary, independent, nonprofit, applied engineering and physical sciences research and development organization with 9 technical divisions.