Drone Inspectors and Iron Hands: A Look at GM’s Plant Technology
At a legacy automaker, bringing in high tech requires both big thinking and tight constraints. Rarely can you build a smart factory from scratch.
More often, you must figure out where, when, and how the latest technology fits into and adds value to an existing operation. It’s a delicate dance that Dan Grieshaber, director of General Motors’ global manufacturing and engineering integration, is very familiar with.
“We simply cannot afford to recapitalize our entire business and just bulldoze everything and start over,” Grieshaber told a roomful of automotive executives recently.
“When you get a greenfield plant to start from scratch, you have an opportunity to link things up very purposefully to be able to take advantage of all the technology.”
But that’s a rarity and a luxury. “The balance of the sites are really, ‘how do we weave this together and integrate this.’”
Grieshaber was part of a panel on manufacturing disruption at the Center for Automotive Research’s annual Management Briefing Seminars, held in late July in Traverse City, Michigan.
While leaders from the likes of Boeing and Toyota de Mexico talked more broadly, Grieshaber delved into the nitty-gritty of how GM uses technology on the plant floor (or in some cases, on the ceiling).
Theirs is a practical approach, one that asks: What sort of technology will improve our business model? How will this transform our plant floors? How will bringing this in solve a specific problem?
Answering these types of questions and integrating new technology into the plant is an interdisciplinary effort—requiring a “blending” of roles: in manufacturing operations, on the shop floor, in manufacturing engineering, and in IT, said Grieshaber.
“Today, we think of these [roles] as distinctly different functions,” he told the group. “In tomorrow’s world, they will be quite merged together.”
At GM, manufacturing high-tech tends to fall into three different areas: improving worker safety, improving flexibility and quality, and using big data to transform the plant floor. Here are some of the highlights:
Worker Safety
Drones
At the Defiance (Michigan) Casting Plant, drones in wire cages are remotely operated to fly to certain areas in the plant‘s interior, performing inspections in places that formerly put humans at risk.
Grieshaber shared a slide of a drone performing a ceiling inspection, eliminating the need for a human operator to “effectively crawl around in the rafters of the plant” with a lift and safety harness.
Another slide showed the drone swooping into an enclosed sand mix vessel for an inspection, “and preventing an operator from having to crawl in there. That’s not an operation I would want to do!” he declared.
The drone takes photos and digitally tags locations, so the operator must only venture in when the inspection reveals that maintenance is needed. “We have the same type of technology on rooftops and outside of buildings,” he elaborated.
Wearables
In the wearables department, harness-like exoskeletons redistribute loads to ease the strain on an operator’s body and provide support, reducing sprains and strains.
Grieshaber held up his arm in an L-shape, miming the rapid raising and lowering of a drill. “If we have an operator on the plant floor who is fastening under a vehicle and doing this all day long...imagine that operation, right? Most of us would be fatigued just holding our hand above our head all day long, let alone performing the task.”
For tasks that require repetitive gripping, an operator-controlled iron hand or “robo-glove” provides additional support and reduces hand and arm fatigue and stress.
“It’s powered by a lithium-ion battery pack and sits in a pack on your back,” said Grieshaber. “There are sensors on the fingertips, and as you grip, it increases your grip pressure. Not to the point where you can crush things, but to where you’re performing an operation of repetitively gripping through an entire work shift.”
Improving Flexibility and Quality
Simulated Training
GM traditionally trains operators by slowing down the line so that they can learn how to build a new product.
“We’re building a prior generation of product while converting to a new generation of product,” said Grieshaber. But that cuts into profitability, so the new approach is to keep the line going at the usual pace and train workers off the line in virtual environments.
“We have standardized work visualization through augmented reality or wearable devices,” he said.
Machine Vision
A machine vision system in the new Flint, Michigan, truck paint shop does a full vehicle paint inspection at line rate, “...which is now at a much higher fidelity than an operator or inspectors would be able to see,” he said.
Collaborative Robots
At the Orion Assembly Plant, in the tire and wheel area, the cobot assists the operator in a task “...that would be inherently difficult from an ergonomics standpoint without some type of lift support,” he said.
Cobots have gotten safer, and these do not require a fenced-in area to protect the humans working nearby.
3D Printing and Rapid Prototyping
GM is not currently printing high-volume parts, but uses 3D printing for “manufacturing aids” to assist the workers, said Grieshaber, “whether it’s a guide pen that’s used to do shock alignment on an underbody carriage, or an ergonomic assist tool or an alignment or gap-setting tool between a fender and a headlamp.
These are things we can make in the plant on the fly to solve immediate business problems, as opposed to having to contract these things out. And they’re sort of penniless to produce because they don’t need to be made out of exotic materials."
Data and Artificial Intelligence
GM views data analytics on a continuum, with data on the left side, and on the right side, “a fully AI world,” said Grieshaber.
“Where you live in between those two bounds in our view depends on the problem you’re trying to solve. Sometimes, staying on the left side and understanding the data you have and being to take action is enough.”
Other situations, like predictive maintenance activities, require more AI. Too much artificial intelligence in the wrong place, and “it’s more things not to work," he cautioned.
Zero Down-Time Technology
GM built this preventative technology for “virtually all” of its body shops with Fanuc, its strategic robot supplier.
“We upload performance information from our robots every night” and run analytics that looks at how the robots are performing compared to their normal operational profile, Grieshaber said.
“And then directions come back to our maintenance departments, that tell us what we need to go work on proactively, in advance of a robot breaking down and then impacting our production.”