Moving from Manual Visual Arc Weld Inspection to Robot Vision

Most components produced by original equipment manufacturers (OEMs) and Tier 1 manufacturers are welded using robots, but the majority of the welds are still manually visually inspected.

AWS Publications | July 21, 2021 | Inspection
Welding Digest ►  Moving from Manual Visual Arc Weld Inspection to Robot Vision

Most components produced by original equipment manufacturers (OEMs) and Tier 1 manufacturers are welded using robots, but the majority of the welds are still manually visually inspected.

Typically, gauges are not used but instead welds are just visually reviewed, and a person makes a subjective go/no-go decision. To assist with this inspection, workmanship or boundary-welded samples are used for comparison purposes. While this approach has been in place since automotive components began being gas metal arc welded in the 1950s, it has resulted in overdesign, excessive repair, and defective parts reaching customers.  

Most of these problems can be overcome proactively by using laser vision seam finding and tracking to adapt to typical joint variability, which is the most common reason for unacceptable welds. However, there is still a requirement to verify weld quality by visual inspection.  

Fully automated robotic arc weld inspection is not only faster and more reliable than manual inspection, but it also provides valuable data that fits well with Industry 4.0 requirements for both automotive and general industry welded components. This data can help predict when a weld is trending toward becoming defective as well as possibly point to what needs to be improved, such as part quality, fixture repeatability, or the welding process itself. In addition, there is the opportunity to reduce the number of cut and etch destructive tests.  

This article is based on more than 35 years of automatic and robotic process management and control experience. It will show the proper steps, tools, and methodology to implement to practically ensure that weldments are produced to consistently meet a defined welding quality level for automotive and other applications. 


Recipe for Success 

The keys to successfully installing an automated weld inspection system are listed below. 

 -Choose a system that is designed for the inspection of arc welds in a high-production arc-welding environment, where the variability and environment are challenging. This system should be based on 3D laser vision scanning technology that utilizes the most precise and rugged 3D laser vision camera. 

-In addition, there must be a seamless and smart interface to the robot, along with associated calibration and verification application software to ensure reliability, and a user-friendly human machine interface to enable quick system installation and monitoring. 

-Then, one needs software that allows for the implementation of proven methodologies and surface map analysis to validate that the vision system is accurate and repeatable.  

-To properly evaluate a manufacturing process, one needs to begin with a process that is in some level of statistical control. This means one needs to measure actual production welds to see how much they vary from the weld that was tested based on the approved welding procedure specification (WPS). The 3D laser scanning system can uniquely provide the ability to establish this welding processes capability. 

Fig 1-Jul-21-2021-02-00-29-30-PM

Methodology to establish welding process capability suitable for successful automated inspection.   


-Once the process capability is proven to be acceptable, then a golden part with the gradient of allowable weld tolerances can be generated, and possibly 3D printed, from the surface map created by the scanning system. This then allows one to ensure the measuring system is properly calibrated, running correctly, and producing the acceptable level of reliability needed.  

-The final step is to test the system in a pilot cell to determine if the allowable differences from the golden part can be detected reliably across normal production weld variability.  



To be successful, one needs to approach automated weld inspection from a welding technology perspective so all the required prerequisites are in place prior to running the robotic inspection system in production. 

There needs to be an effort to update and harmonize the welding quality standards of OEMs, Tier 1s, and American Welding Society (AWS), which have historically been used for manual human inspection, to ensure these standards include what is required for successful weld inspection using 3D robot vision systems.  

Automated weld inspection can offer many benefits to both automotive and general industry OEMs and Tier suppliers for the improvement of weld quality, overall productivity, and environmental friendliness. If properly applied, robot vision systems, possibly involving the latest collaborative robots, are very reliable and can replace human inspection, which is costly, tedious, and inconsistent. 


This article was written by  Jean-Paul Boillot, Jeffrey Noruk, and Sébastien Latour for the American Welding Society.