Ask the “personal assistant” that sits on a shelf in your living room for the location of the closest pizza place. The assistant responds with three choices. Enter your preference into your phone and, once you are in the car, a navigation app guides you there. While your devices gave you information using artificial intelligence (AI), you made the decisions. But that’s not always the case with AI. Beyond the conveniences it brings, AI in its many different forms has applications across our lives.
As part of the effort to explore the potential AI offers, faculty at Penn State Altoona have joined other universities and businesses in the AI Innovation Consortium, “working to promote cooperation between companies working on AI applications and the universities,” says Jungwoo Ryoo, head of Penn State Altoona's Division of Business, Engineering, and Information Sciences and Technology and professor of information sciences and technology. “One of the goals is to set up incubators at the universities involved to illustrate how AI can be used in companies.”
Of great interest to businesses is “machine learning,” a form of AI where a machine can actually learn something and make decisions based on what it has learned. Bruce Muller, senior instructor in engineering at Penn State Altoona, teaches EMET 430, an advanced automation course. He says, “I try to stay up on newer things,” he says. “My area is manufacturing.” At a conference in Toronto, “I started communication with a company involved in manufacturing. My goal has been to try to integrate a number of labs into my EMET 430 course.” Out of this relationship, Muller was able to create a lab for his students in Altoona that utilized machine learning.
Ian Core, Industrial IOT Specialist at V-Soft; Konrad Konarski, Artificial Intelligence and IOT Practice Head at V-Soft and Chair and Director of Operations at AI Innovation Consortium; Bruce Muller, and Jungwoo Ryoo, with the 2020 Global Fellows at Penn State Altoona
“The initial foray was a pipe-threading project,” he continues. Inspecting the threading in newly manufactured pipes has always required a human eye. “What the company wanted was a way to use machine learning to see if the threading was right—a chip, a dent, etc.,” Muller says. “We had a camera system and turntable so you can put the short pipe on the turntable. We started seeing what that was like.” As with so many other things, the project was disrupted by the pandemic, so the students didn’t get as far as anyone wanted.
Muller was in no way deterred. In all, he has done three labs with his students that involved machine learning. Another company manufactures refrigerator back panels, requiring “top dies and bottom dies to do the bending and punching as needed. Workers arrange the dies,” he explains. If they don’t set the dies correctly, it could be “an $8,000 mistake. What the company wanted was to use machine learning so that a computer could look at the dies and see where the punches were and say, in effect, ‘they’re all where they need to be, go for it.’”
In the lab, the students didn’t get to that stage, but it was still a success. “They took pictures and, using software already on their computers, they told the computer what a punch looks like,” Muller says. “The computer learned. They annotated the pictures so that the computer could see what they did and fed that into an algorithm to train an AI learning model to recognize punches on a die.” Of course, the final step was testing it: “They fed in a bunch of pictures of dies and the computer got 100%.”
(Left to right) Tammy Simonetti, VP, National & Strategic Accounts at V-Soft Consulting Group, Inc., the primary sponsor of the AI Innovation Consortium; Ian Core, Industrial IOT Specialist at V-Soft; Jungwoo Ryoo, Bruce Muller, Chris Martin, and Corey Griffin, Associate Dean for Research at Penn State Altoona.
Ryoo recognizes the benefits of bringing companies’ problems into the classroom and applying “emerging technologies” such as AI. Everyone wins. He says Muller teaching “how AI can be applied to industry problems” means students learn about real-life industry problems. “The more students know about the technologies now, the better.” Another inter-academic benefit: “This is a great example in terms of collaboration in IT. Engineering needs IT more and more.”
Another Penn State Altoona faculty member interested in the benefits of machine learning is Chris Martin, associate professor of mechanical engineering. “I have a background in manufacturing,” he says, which gives him a valuable perspective on the potential for AI and industry. He calls himself a “novice in the world of AI,” but he also recognizes that his own present research “is going down an AI route.” Fortunately, “machine learning has progressed so far you can be a novice and still do great stuff.”
Martin sees the benefit in connecting industry and universities. “A company may be really good at AI. Another company may be really good at manufacturing, but they don’t realize they need AI.” Consortiums can bring those companies together with academics who, Martin says, “are really good at taking deep dives at problems. We need all these players at the table.” And despite his self-declared “novice” status he too has much to offer: “My work is to be part of the conversation in our collaboration.”
To continue to participate in that conversation, Ryoo, Muller, and Martin will be attending Evolve 2021, the annual meeting of the AI Innovation Consortium, in August. Ryoo and Muller will be part of a panel discussion on “Keeping up with Artificial Intelligence in Academia,” and Martin will be a panelist for the “AI and Manufacturing” session.