How COVID-19 Drove Automation Advances

Image Source: alexlmx/Gettyimages.com

By David Freedman for Mouser Electronics

Published May 9, 2022

Editor’s Note: This article was written based on the author’s interviews with multiple experts in the automation industry.

In 2021, Westec Plastics put a giant “We’re hiring!” banner in front of its Livermore, California, building; paid for listings in the major online hiring sites; and increased starting pay, which was already far above minimum wage, by 24%. None of it helped; for more than a year, Westec has been short operators on all three shifts, and the company lost business as a result.

But now Westec has gained that business back and more. The 100-employee company recently installed three “cobots”—robots that work alongside human employees—that have assumed some of the simpler tasks (Figure 1). Westec is still trying to hire more people, but because three cobots do the jobs of six people, the person-power shortage is much less severe. “It’s working out great,” says Tammy Barras, Westec’s president. “We’re placing an order for two more cobots, and we plan to keep building up our robot staff.”

The drive to automate began long before the pandemic, but the labor shortages and supply-chain disruptions that emerged over the past two years accelerated what was already a steady trend. Companies that may have been delaying or moving slowly toward automation suddenly had plenty of reasons to take the leap. “They no longer had to wonder if it was feasible for them to automate,” says Thorsten Wuest, associate professor for smart manufacturing at West Virginia University. “They didn’t really have a choice.”

Companies like Westec that launched or upgraded automation during the pandemic—and that saw a quick return on investment—are likely to add more of it, according to Wuest. “The new question is how to best take advantage of automation now that operations are getting back to normal,” he explains. Those advantages are growing, he adds, largely because robots and other types of manufacturing automation are becoming smarter, more flexible, and less costly.

Figure 1: Smaller manufacturers like Westec Plastics have mitigated labor shortages by installing low-cost self-teaching robots capable of working alongside employees. (Source: Westec Plastics)

A Big Payoff

A November 2021 study by SnapLogic and Cebr revealed that U.S. companies realized an average increase of 7% annual revenue rate within three months of investing in automation. At the same time, those companies increased employment by 7%. The idea that automation would lead to more jobs sounds counterintuitive, but it actually makes sense, according to Jordan Kretchmer, chief executive officer of Rapid Robotics, a San Francisco-based company that provides robots to manufacturing companies including Westec. “All of the studies in the past few years show that a decrease of 10% in the cost of labor as a fraction of product cost leads to an increase in margins of 20%,” says Kretchmer. “And bigger margins mean more potential to hire and upskill employees, and to keep employees longer.”

The equation has swung even more strongly in favor of automation over the past few years, with manufacturing wages increasing 10% since early 2020, according to the U.S. Bureau of Labor Statistics. And hiring people to perform less-than-pleasant jobs is difficult. “People don’t always want to sit in front of a machine and get sprayed by metal shavings all day long,” notes Kretchmer.

Meanwhile, companies have been struggling with the supply-chain disruptions that have roiled business around the world. “One week we’d have a vendor overseas building tools for us, and the next week they’d be shut down,” says Barras. “Now our customers want us to have all our tooling done domestically to avoid problems.” But the shift to demand for all-domestic manufacturing has only exacerbated the pressure on hiring.

Smarter Automation

Automation vendors are adding artificial intelligence (AI), machine learning (ML), and other advanced software to their equipment to solve these and other problems for manufacturing companies. “There has been a big acceleration in adding intelligent tools on top of the more basic type of automation that companies have been focusing on for the past 20 years,” says Heath Stephens, digitalization leader at Hargrove Controls and Automation, an automation systems integrator in Mobile, Alabama.

Stephens notes that many manufacturers long ago installed equipment that records data from sensors tracking temperature, pressure, flow, and dozens of other potential variables. But most companies haven’t been able to do much with the data, because it would take a team of highly trained analysts to sift through them and perform the complex calculations needed to improve automated operation.

ML, however, can sort the data in real-time and automatically adjust multiple controls in a manufacturing process to improve results—for example, increasing the purity of a chemical that’s being produced, reducing energy costs by lowering the temperature of a process without compromising the output, or minimizing material waste to boost sustainability (Figure 2). “When you’re dealing with a process that has 10 parameters, a person isn’t going to be able to adjust them all to find the perfect sweet spot,” says Stephens. “But with an ML program, you just tell it what the goals are, and it figures out how to get there.”

Besides improving output and lowering costs, the new automation intelligence also ups the predictive maintenance game. Stephens notes that conventional predictions use simple sensing techniques like increased vibration to provide warning only a few hours prior to failure. But now ML programs, many of which are cloud-based, can spot complex patterns in multiple sensors and offer failure warnings at least one week in advance—plenty of time to order needed parts and plan maintenance during scheduled downtime, saving costly shutdowns. “The added intelligence isn’t about reducing personnel; it’s about improving efficiency and quality,” remarks Stephens.

Wuest explains that adding AI-based capabilities to manufacturing processes and robotics results in “cognitive automation,” and notes that what makes it most powerful is the way it’s being integrated into conventional “physical automation.” “The machine learning, the analytics, the vision systems, the sensors, all of it now go hand in hand with robotics and other machines,” says Wuest.

Figure 2: Manufacturing automation incorporates AI-based programs that oversee and adjust processes to increase output, lower costs, and improve sustainability. (Source: Hargrove Controls and Automation)

More for Less

Adding robotics has long been a daunting proposition for smaller manufacturing companies owing to the cost and complexity of programming and setup. According to Kretchmer, it can cost hundreds of thousands or even millions of dollars to set up robots on an assembly line. Although such an investment pays off for a big company that’s going to produce the same part in the same way for several years, it usually doesn’t work for smaller companies. “Ninety-eight percent of U.S. manufacturers have assembly lines that constantly change from order to order,” says Kretchmer. When it costs $100,000 to reprogram a robot with every assembly line change, the investment doesn’t make sense.

To bypass that problem, robotics companies are now building ML-based capabilities for self-programming into robots. For example, Rapid Robotics’ “Rapid Machine Operator” robot views a computer-aided design diagram of a component and what task needs to be performed. Then the robot uses its four cameras to analyze its workspace and determines on its own how best to complete the task to 0.1mm accuracy—for example, feeding a part into a machine and then removing it, operating the machine itself, or welding. If a component gets out of position, the robot accommodates that variation on the fly.

Because AI-equipped robots are so easy to set up, they can be moved between assembly lines or can continue working as an assembly line is reconfigured for a new order. “You can roll our robot over to a new machine, and in thirty seconds it will start production, just like a human would,” explains Kretchmer. In contrast to the enormous costs of buying and setting up a conventional robot arm, the Rapid Machine Operator is priced as a $2,100-per-month subscription.

Wuest notes that as cognitive automation becomes more capable, it will start taking over higher-skilled jobs. He doesn’t doubt that the benefits of this increasingly sophisticated automation will continue to create more and better opportunities for human employees. The risk, he says, is that many of the people whose current jobs are assumed by automated equipment won’t have the skills needed to advance to the new jobs that are created.

To address that problem, schools, companies, and governments have to put more effort into upskilling employees, contends Wuest, so that workers are ready to take advantage of the new opportunities that automation-fueled productivity will create. “Combining automation with upskilled employees is the best of both worlds,” he says.