MIG, TIG, stick, or laser—producing high-quality weld joints requires considerable knowledge, a steady hand, and plenty of practice. Are robots up to the task?
From flipping hamburgers to helping Grandma around the house, it’s no secret that robots are getting good at tasks that have long been considered off-limits. Nowhere is this truer than in the manufacturing industry, where a chronic labour shortage is forcing machine shops, plastic injection moulding houses, and sheet metal fabricators to automate like never before.
Machine tending, deburring, assembly, and packaging: these are just a few of the typical manufacturing processes where robots and their collaborative cousins have begun playing a leading role over recent years. And despite its relative difficulty and need for nearly flawless hand-eye coordination, welding is quickly becoming another operation where robots (almost) rule.
Checking the boxes
Several caveats exist, however. “First off, the parts must have consistent dimensions and fit securely within the tooling, which should be designed to accommodate automation and have plenty of access to the features being welded. If not, there must be a way to move either the robot or the part to ensure a high-quality weld.”
That’s according to Mark Scherler, general manager at FANUC America Corporation, who explains that accessibility is relative to the size of the workpiece and the location of its various features—for small parts, the welding process is usually straightforward, but for larger ones, achieving a successful weld can be more challenging.
Lastly, and perhaps most challenging of all, is the need for a skilled individual who can both program a robot and understand the nuances of creating a good weld. “Workers typically fall into one of two camps—welders and robot programmers—and it’s rare to find those who possess both skill sets,” says Scherler. “To address this issue, you have two options: either teach a robot programmer the art of welding, or find a welder willing to learn how to program robots. It’s only by bridging this expertise gap that you can ensure success in automated welding processes.”
In most cases, Scherler would hire the latter. Given programming software’s ease of use, doing so will generally take less time and produce better results than attempting to teach a non-welder the intricacies of what is clearly a complex process. “Welders have an innate understanding of variables like torch orientation and gap distance, all of which helps to make the programming process easier and more effective.”
Humans can also compensate for poor fit-ups and parts that might not sit securely in the fixture, says Scherler, whereas a robot will need advanced capabilities like a laser-based weld seam tracker to adjust for unexpected part variations. And even with these systems, automation works best when components have minimal variation, joints designed for weldability, and high-quality workholding.
Safety is another concern. As most would-be automation owners have found, the industry offers two distinct types of robotic arm—the classic industrial robot that we’ve been seeing on automotive assembly lines for decades, and the newer “force- and speed-limited” versions dubbed collaborative robots (cobots) for their ability to work alongside humans with minimal to no cages or guarding.
FANUC offers both, but as Scherler points out, the latter is enjoying a warm welcome in many job shops due to its simplified setup and programming requirements. Joe Campbell, senior manager for strategic marketing and applications development at Universal Robots USA Inc. (UR), seconds this viewpoint. “During my decades in the automation industry, I’ve walked into any number of high-mix, low-volume shops where the people on the floor state they can weld a job faster than we can program it. There was some truth to that in the past, but it’s rarely the case anymore.”
Nor is it the case that cobots lack the precision and rigidity needed for welding. The last few years have brought significant technical improvements to this space, which is why UR, FANUC, and every other leading robotics manufacturer not only offer a collaborative product line, but are beginning to promote their use in this final frontier of automated manufacturing.
Back to the future
Says Campbell, “I once worked for a large automotive engineering company in Detroit that would put an entire team of robot technicians into a plant, programming the robots, simulating the assembly line, tweaking and adjusting everything until it was just right. The process took months, but it was okay because they would build a hundred thousand cars a year for the next decade. That’s no longer acceptable, which is why we’ve designed a cobot that even a moderately skilled person can set up within minutes for practically any application, welding included.”
Despite this apparent simplicity, Campbell points out many of the same caveats as his counterpart at FANUC—namely, the need for consistent part dimensions, stable, well-designed workholding, and preferably, a skilled welder with the desire and aptitude to expand their vocational horizons.
People are listening. Welding has become the fastest-growing application segment at UR, and low-volume production accounts for much of it. “Whatever the company and whatever they produce, everyone from Mom and Pop shops to Fortune 500 manufacturers are putting in cobots for their smaller lot sizes, even down to 100-piece part runs,” he says. “Doing so increases part quality and throughput while freeing up skilled workers for more challenging tasks.”
And unlike Campbell’s “back in the day” war stories, where programming time was best measured in days or weeks and the implementation and training of a new robotic system took months, cobot owners are making parts within hours of their little droid’s arrival. “Time is money for any size company, but it’s the small shops especially who can’t afford the downtime; they need to get into production quickly. That’s just as true for welding as it is for machine tending, assembly, grinding, deburring—name the manufacturing operation; chances are good that we can automate it.” SMT