Home » NSF CAREER Award advancing UK researcher’s work

NSF CAREER Award advancing UK researcher’s work

Integrating AI in manufacturing

LEXINGTON, Ky. — As industry experts navigate the excitement and concerns surrounding the use of artificial intelligence (AI), one University of Kentucky researcher is exploring its potential integration into manufacturing through a National Science Foundation (NSF) Faculty Early Career Development (CAREER) Award.

Peng “Edward” Wang, Ph.D., is the Robley D. Evans Faculty Fellow and an assistant professor with a joint appointment in the Electrical and Computer Engineering and Mechanical and Aerospace Engineering departments in the UK Stanley and Karen Pigman College of Engineering. He joined the university in 2019.

“My NSF CAREER Award will focus on developing the first large-scale generalizable machine learning model specifically tailored for manufacturing applications like welding. We aspire to develop a groundbreaking AI machine learning solution that can be likened to the manufacturing counterpart of ChatGPT,” said Wang.

Photo by Jeremy Blackburn | Research Communications

ChatGPT is an AI language model that processes language to generate relevant responses, but its limitations include the inability to process image-based manufacturing data.

The NSF will support Wang with $567,930 over five years for his research on AI and machine learning (ML), where he will utilize real-world production data to develop a system that can be applied broadly to manufacturing systems to improve autonomy, manufacturing efficiency, quality control and sustainability.

This award is one of the “most prestigious awards in support of the early career-development activities of teacher-scholars who most effectively integrate education and research within the context of their organization’s mission,” according to the NSF website.

“I have been working in AI and ML, developing solutions specifically tailored for manufacturing applications, for several years,” said Wang. “However, it has come to my attention that most AI/ML solutions are currently being developed in controlled laboratory environments. Significant disparities exist between the lab data and the real-world production data obtained from manufacturing plants.”

The disparities in the data are visible in areas like data variety, quality and labeling. For researchers like Wang, this prevents applying these AI/ML solutions to practical manufacturing settings.

Over the next five years, Wang and his team will work to address these challenges and bridge the gap between AI/ML research in controlled environments and real-world manufacturing plants. Crucial to that work is the access to complex, real-world manufacturing plant data.

“I think all people working in the field of AI and ML know that the truth is, if you don’t have data, you have nothing to develop a successful model,” said Wang.

The research team is partnering with industry powerhouses General Motors (GM) and General Electric (GE) to provide the data necessary for this work. GM has shared its welding plant and laboratory data, which is a necessary foundation for Wang’s project.

“Collaborating with GM and GE not only provides us with invaluable resources but also offers us the opportunity to test and refine our AI/ML models in real manufacturing environments,” said Wang. “By working closely with these industry leaders, we can ensure that our solutions are robust, practical and aligned with the manufacturing sector’s needs.”

Wang will also collaborate with the Kentucky Association of Manufacturers (KAM) to share knowledge and guide small and medium manufacturers on transitioning from traditional manufacturing practices to a smart manufacturing model that can suit their unique needs.

This NSF-funded project will also build a partnership with the Math Science & Technology Center (MSTC) at Paul Laurence Dunbar High School in Lexington. Up to four students can work on a specifically designed research project, such as utilizing or developing the intelligent robotic welding module for mechanical part design with specific characteristics.

Wang recognizes this work could pragmatically change the future of manufacturing with an envisioned model applicable to diverse materials, machines and production lines. It could also accommodate a range of downstream application tasks in areas like product quality inspection, defect detection and even process optimization and control.

“To me, the NSF CAREER Award is not only a recognition but, more importantly, it provides me an opportunity to work on a real-world challenge,” said Wang. “So if I can be very successful on this project, our group will develop the first successful generalizable machine learning solution for manufacturing that will be very meaningful.”

Click here for more Kentucky Business News.
Click here to Advertise.