America’s manufacturing resurgence will be powered by these robots

Once a luxury reserved for big manufacturers, smaller, smarter, more flexible and less expensive “cobots"—collaborative robots—are bringing automation to every fabricator, no matter the size. These robots aren’t just nice to have. The slow, fragile recovery of American goods production wouldn’t be possible without them.

The number of U.S. companies that make physical things reached a low point in 2014 and has grown since then. Yet they are trapped in a never-ending labor shortage as skilled workers age out, and young people fail to take their place.

China has become the de facto manufacturer of the world’s goods, owing not only to its enormous population of engineers, technicians and machinists but also its 2-million-plus army of industrial robots. Now the U.S. is attempting to claw back some of those contracts—a process called “reshoring"—and robots can in some cases quadruple worker output.

There’s always a danger that companies will over-automate, a trap that has sent many a firm into bankruptcy when recession kicked in, demand dried up and equipment weighed on balance sheets. There’s also a hazard in thinking robots can do everything—adding them generally requires finding humans to work alongside them.

But for now at least, the push to bring manufacturing back to the U.S., and the demand for industrial goods to power America’s AI-fueled economy, are driving automation adoption and innovation.

Here come the cobots

“Automation is key to reshoring, plain and simple," says Greg LeFevre, CEO and president of Raymath, a metal fabrication company based in Troy, Ohio.

When he bought Raymath in 2019, this maker of food-grade machines, earthmover parts, data-center server cabinets and more was nearly four decades old and had almost no automation. Two years later, LeFevre began adding cobots, a kind of lighter, less expensive, easier-to-program industrial robot that’s safe to use around humans. Now he has 13.

Per-worker productivity at his facility has soared, with individual workers producing as much as four times as many pieces per day as they could before. This makes Raymath more competitive, and has helped the company triple its revenue. LeFevre has also been able to slightly expand the company’s workforce, from about 130 to 145 workers. Humans have gone from doing all the grinding, welding, machine-operating and parts-moving to overseeing and collaborating with robots who take on the bulk of the tasks.

These robots have become radically easier to program over the past decade, and now people can use a simple tablet interface to instruct them to perform specific sequences of actions. Programming the older robots common in automotive factories since the 1960s took years of training.

Dozens of companies in the U.S. now offer manufacturers the kinds of specialized welding cobots that LeFevre uses, says Josh Pawley, co-founder of Loveland, Colo.-based Vectis Automation. Founded in 2019, Vectis was one of the earliest U.S. companies to assemble welding cobots. (LeFevre’s shop uses robotic welders from a competitor, THG Automation, based in Indianapolis.)

These metalworking cobots are part of a broader trend in robotics: Specialized robots that use sensors to safely navigate human environments. They can cope with more variability than previous industrial robots, which had no sensing abilities. This has been essential to the rise of Amazon and its superfast fulfillment, and now it’s coming to manufacturing.

Manufacturing cobots are a far cry from the humanoid robots which have garnered so much attention of late. Elon Musk has staked a compensation package that could be worth a trillion dollars on his promise that Tesla will produce a million Optimus robots within a decade.

Humanoid robots lack strength, dexterity and specialized appendages—like welders or grinding wheels—but startups building them are betting they compensate through versatility. Perhaps they might unpack boxes in the morning, then pack other boxes in the afternoon. Someday they could, in theory, pick up a welder or a grinding tool, but making them dexterous enough is a huge challenge. Musk recently said that for Optimus, “the hands, inclusive of the forearm, are a majority of the engineering difficulty of the entire robot."

David vs. Goliath

China is indisputably the leader in high-volume manufacturing, and companies that want the biggest volumes of manufactured parts for the lowest possible price continue to send work there. And though many U.S. manufacturers can’t match their Chinese peers in volume, they are competing by using automation to tackle smaller batches of goods under tight deadlines.

Clients that wouldn’t have thought of manufacturing in the U.S. are now asking how to reshore the making of critical parts, says Jack Callender, president of Caltech Manufacturing in Bucks County, Pa.

Each metal part his shop makes—for medical devices, vehicles, scientific and industrial systems and more—can cost more than the quoted price for a piece from China. But customers who contract with him are choosing to avoid some of the challenges of overseas manufacturing, he adds.

Callender’s shop employs more than 80 people and, in the span of just a few years, has added a half-dozen welding robots alongside a menagerie of other automation systems. Its laser cutters can load their own sheets of metal, like giant printers with automated feeder trays. Like LeFevre’s shop, Caltech Manufacturing has seen an increase in per-worker productivity of between two and four times, while expanding its workforce, he adds.

In just the past few years, the robots Callender is using have become smarter. They now incorporate touch sensors so they can verify that they are welding in the right place. Callender is also considering systems with computer vision for more versatility.

U.S. investors are making bets on yet-to-be-proven technologies that could yield even more capable robotics.

Bay Area-based Dyna Robotics is making a “foundation model" for manufacturing robots. This means they learn to perform tasks, instead of being programmed to do them. Today, the startup’s robot can fold dinner towels, but soon, the firm will partner with real-world factories to expand that skill set, says Lindon Gao, the company’s CEO. His company announced last month it had raised $120 million from investors including Nvidia and Amazon.

It’s a long-term play. Making these learning robots reliable enough for an industrial environment is next to impossible at present, says Christopher Müller, director of the statistical department at the International Federation of Robotics.

That’s not stopping Gao, who hopes to deploy them within a few years. Once robots can be set to work with a quick demo and some verbal commands, the country that can muster the most robots—rather than the most workers—wins.

Write to Christopher Mims at christopher.mims@wsj.com