Part 2 of an occasional series on automation, robots and AI. Read Part 1.
With nary a human in sight, “Kong,” an aptly named monster of a robot, lifts entire truck frames weighing thousands of pounds at the Kentucky Truck Plant.
Nearby, thanks to a 2016 infusion of $1.3 billion by Ford, 1,000 robots twist and turn as they effortlessly lift heavy auto parts, such as door frames and hoods.
Indeed, entire sections of the six-million-square-foot plant in eastern Louisville require essentially no humans.
Until they do.
Last year, the company pumped another $900 million into the plant to prepare for the new Lincoln Navigator, and now humans and robots are working side by side.
It’s an evolution of sorts: First robots worked separately from humans. Now robots are working next to humans. It begs the question: Will robots be working instead of humans soon?
Joe Hinrichs, Ford’s president of the Americas, doesn’t think so, at least not in manufacturing plants.
“There’s a human element in this that’s really important, and I don’t see that going away any time soon,” he said recently during a tour of KTP.
As of now, humans still remain vital to the plant’s operations: That $1.3 billion investment that brought in 1,000 robots? It also required the addition of 2,000 humans. And the $900 million investment last year retained 1,000 (human) jobs.
“There’s a lot of robots out there,” Hinrichs said. “There’s also 8,400 people … I don’t see that changing in the automotive business any time soon.”
Hinrichs said that robots cannot compete with humans — at least not yet — when it comes to dealing with the unexpected.
“There’s a significant amount of problem-solving that goes on every day, and that’s real time. And that can’t be done by robots or automated systems,” he said.
If something breaks down, humans have to intervene to fix it. If a machine acts up, humans have to figure out why. If a part comes in differently than previously, humans have to determine why and find a solution.
Robots are used primarily for tasks that humans cannot — or, for safety reasons, should not — perform.
Ford’s human employees are “not just laborers,” Hinrichs said.
“The eyes and ears and the experiences of people out there help us,” he said. “They help us produce a better product, help us improve. They help us identify problems.”
Automation brings jobs to GE Appliances
About 15 miles southwest, in GEA’s building AP5, a refrigerator plant, the situation is similar.
A big robot there lifts refrigerator shells that weigh 280 pounds. It does so in a fenced-in area, to keep humans away, for their safety. Nearby, machines feed presses with flat sheets of stainless steel for the refrigerators’ shiny exterior. Automation means that humans don’t have to risk grave injuries by sticking their hands and arms into the presses, which exert a force of 330 tons.
Rob Willy, GEA’s executive director of advanced manufacturing, said the company uses robots primarily to reduce “the four Ds:” jobs that are dirty, dull, difficult or dangerous.
And, he said, in the last few years, technological advances have prompted the company to bring in-house processes that it used to outsource: The refrigerators’ stainless steel exterior, which the company now presses in-house, used to be handled by a supplier. The bottom pan of the refrigerators’ freezer, which the company also makes on site, would have been outsourced as recently as three, four years ago.
A lot of the injection molding used to be shipped to the Louisville campus, but is now being done — largely by machines — in AP5. GE Appliances now has the largest injection molding shop in Kentucky, said Willy, who has been with the company for more than 20 years.
GE Appliances began focusing on vertical integration about a decade ago, he said, because it learned that keeping as much of the production in-house as possible provides valuable insights about design, quality and efficiency.
Willy’s team consists of about 135 people in the U.S., and a handful overseas. It designs, builds and introduces technology to keep the company at the leading edge. Willy said the team makes sure that the workers have a factory that enables them and the company to be successful.
Having designers on-site, in the midst of production, enables them to observe challenges and interact with production workers firsthand so that teams of both can immediately improve the designs.
And those improvements — including operator feedback — are vital to making the production as efficient as possible. Initial designs almost always harbor some inefficiencies, Willy said, which also presents a hurdle to automation: “What you don’t want to do is automate waste,” he said.
Robots may excel at picking up an item from a fixed location and dropping it off in another fixed location, but automating the assembly line remains tricky, because a lot of the work requires a variety of complex movements.
When a refrigerator shell approaches on the moving line, workers have to first read the manifest pinned to the object to determine which parts bag they need to grab. Installing some of the components correctly also requires visual inspections and tactile and audio feedback: Is the part in the right spot? Did it snap in correctly? Did it make a clicking noise?
And, Willy said, the workers are dealing with components that, depending on the model they’re building, are not always in the same place and do not always go into the exact same spot. Plus, they’re affixing components to a moving target.
“That’s very difficult for a robot,” he said.
Hinrichs agreed. A lot of processes have been automated already, and that probably will continue, but there’s always a trade-off. If something doesn’t go as expected, a line populated by robots shuts down. Humans, on the other hand, can react.
Uptime is critical. Even if a robot performs a task flawlessly 99 percent of the time, but stops working 1 percent of the time, a line with 10 of those robots is going to shut down 10 percent of time.
Making improvements to processes still requires people, Hinrichs said. Machines cannot yet watch themselves to create new, better processes.
Willy said that a lot of products, including refrigerators continually are getting more complex, as consumer demand that they dispense water, crush ice and make coffee.
The left door of a fridge with a water/ice dispenser has more parts than an entire dishwasher, he said.
While bringing back injection molding and stamping operations has increased production efficiency and product quality and flexibility, it has brought its own set of challenges for both GEA: finding enough employees with skills to program, maintain and repair an ever more complex set of machinery. That’s especially true now, with a national unemployment rate below 4 percent. Many other local businesses are dealing with the same challenges.
“Good people aren’t looking for jobs right now,” Willy said.
For GEA, that has meant equipping current employees with more skills, he said.
Data, mobility, visualization
Both Willy and Hinrichs said that the companies increasingly are relying on data collection and analysis to prevent downtime by identifying tasks with which employees are struggling and by predicting when machines are about to fail.
At GEA, every piece of machinery is equipped with programmable logic controllers, and while the company has been collecting data for a few years, it only recently has been able to present the information in a usable form: with graphics and on mobile devices.
Willy can use his phone anywhere in the world and find out how close a particular production line is to operating at maximum. The system can even let him know which employee on a line may be slowing down production more frequently than others — and with which task the employee is struggling.
On the refrigerator line, one of the tools used by the workers tightens screws. The tool can be set to detect whether it faces enough resistance from the screw. If it doesn’t, because the operator may have removed the tool from the screw too soon, the line may stop. If that happens repeatedly, a supervisor’s mobile device can alert him to the problem so that he can talk to the operator. If multiple operators struggle with the same task, it could be a sign for the design team to change some processes on the line.
Willy said employees used to share information about production stoppages and other problems in end-of-day meetings, and issues sometimes weren’t addressed even then. Now the information is available right away and enables supervisors to respond with much greater velocity than before.
Ford, too has improved its data collection and the format in which it is displayed to allow employees to more easily diagnose and address problems, Hinrichs said. Large video displays at KTP provide detailed information about each manufacturing line, enabling supervisors to detect problems immediately, even if the line is far away and out of their sight.
The next step will be to use the data collected by machines to have them predict failures. Willy said sensors on various parts are collecting data about vibration and temperature changes and will be able to predict when a parts failure is about to happen. That will allow workers to inspect the parts and replace them if necessary, preventing unexpected — and costly — production interruptions.
Ford has been building enormous data centers to collect, store and analyze the information. The automaker announced in March 2017 that it was spending $200 million to build a second advanced data center in Michigan, “as the company expects its data usage to increase 1,000 percent, driven by manufacturing and business needs and new mobility services, such as more connected, autonomous and electrified vehicles.”
Willy said that data that Ford and other automakers are collecting from autonomous vehicles also will be applied to material handling in manufacturing and logistics plants, as robots will use the information to figure out the most efficient ways in which to pick up and distribute components.
So far, the increased use of robots and the resulting efficiencies and flexibility, have resulted in record production numbers in GE Appliances’ building AP5.
“This year, Appliance Park is producing more (refrigerators with) bottom freezers than ever before,” Willy said. “We have two shifts making as much product as we can.”
Join Insider Louisville July 24 for Rise of the Robots, a discussion with local experts, including Willy, on the ways in which robots and AI are affecting the local economy.