Leveraging his expertise in image processing, sensors, and machine learning, Vincent Paquit is devising a control system for additive manufacturing to produce 3D-printed parts that function as well as conventionally produced objects.
Paquit’s research sits at the junction of manufacturing technology and materials science, drawing on a broad spectrum of techniques—data collection, data visualization, data processing, and artificial intelligence (AI)—to better understand how materials react during the printing process. He gathers and analyzes data to identify the occurrence of defects and to predict material properties. Those insights are then used to create algorithms to guide computer-driven 3D printers.
“We want to translate our understanding of 3D-printed material properties into hardware and software development for these machines,” said Paquit, data analytics team lead at the Manufacturing Demonstration Facility (MDF) at Oak Ridge National Laboratory (ORNL). “The goal is to develop a printing scheme heavily driven by the fundamental science of how these materials behave, rather than bounded by empirical engineering and computing choices.”
He and his colleagues are determining the parameters required to 3D-print a part that can be validated and will meet certification for end-use. “Our goal is to complement existing modeling and simulation efforts with a data-driven approach to qualification and certification of additive manufacturing processes,” he explains.
The use of traditional non-destructive techniques such as computed tomography scanners to view the internal features of individual parts on a production line is cost-prohibitive, the researcher noted. “Instead, we want to develop a system using sensors embedded in the machinery and custom algorithms to identify any anomalies in real-time that could affect an object’s material or mechanical properties,” Paquit said. “If a hiccup is observed, we want the system to automatically determine whether any resulting defects will affect the function of the final product.”
Paquit added that for industry to absorb this type of technology, “we will need something compact, easy to implement, low-cost and very reliable. Although the MDF’s research requires specialized tools, in the end industry will require a process as simple as using cell phone to both take a picture and do follow-up processing.”
To get to that point, Paquit and his colleagues are going through a highly complex process using big science tools at the lab. MDF researchers have used specialized microscopy and neutrons, for instance, to explore the material properties of printed objects. The next stop is high-performance computing: “With an ever-increasing data science portion and the need for AI, developing the next generation 3D printers will require a lot of data-crunching,” he said.
Paquit came to ORNL as a doctoral student in the summer of 2004. His first task in the Imaging, Signals, and Machine Learning (ISML) Group had biomedical applications: developing a non-invasive technique to examine blood vessels under the skin and identify the best sites for catheter insertion. It was part of a larger project to create a method for remotely treating wounded soldiers in the field when medical personnel are not present.
“If you go to an emergency room, having an intravenous catheter inserted is one of the first things done so you can receive fluids and medicine. But people are different; their veins may not be readily apparent due to their physiology or whether they’re hydrated,” he said.
The approach developed by Paquit and his colleagues—and which became the subject of his dissertation—involved the use of different wavelengths of light to enhance blood vessel contrast so their location and depth can be estimated.
Paquit credits his math teacher mother and a landscape architect father who designed golf courses for his natural interest in math and science while growing up in France. “I was always seeing math and geometry and drawings back then, and I chose a concentration on math in high school,” Paquit noted.
Paquit initially intended to pursue a career in medicine but steered back to math and physics once he began college. Paquit earned his bachelor’s, master’s and doctoral degrees in electrical engineering from the University of Burgundy in France. He was also a research associate in the technology transfer office at the university, where he gained experience in collaborating with industry that serves him well at the MDF, a U.S. Department of Energy (DOE) user facility available to industrial partners who want to explore better manufacturing methods with ORNL scientists and engineers.
A passion for new discoveries
Paquit’s time at the MDF began when researchers there sought assistance from the ISML Group on imaging processing for the parts they were printing. As the MDF’s research has expanded, so has his analytics work.
“Being part of this project that can have a wide impact, the excitement of collaboration and constant new challenges, is what drives me forward,” Paquit said. “It’s extremely motivating to be in an environment where you have diverse expertise that drives rich interactions on a variety of topics.”
Away from the lab, Paquit enjoys time with his family—wife Sophie Voisin, a researcher in the Geographic Information Science and Technology Group at ORNL—and two young children. He also enjoys woodworking. “I like doing manual work in my spare time,” he said. “It’s a way to relax and at the same time create something nice for our home.”
Paquit has crafted office furniture, a fireplace mantle, and bedroom furniture for both his children. “I’m having fun, but the downside is my two-car garage is now a one-car garage with all the equipment I’ve acquired,” he said.
Paquit’s passion for new discoveries is well suited to his research work. “I like being able to say I’m happy to wake up in the morning and go do my work at the lab. I’m learning new things every day and seeing things I’ve never seen before,” he said.
“The constant refreshing of the science here means that there’s always a new challenge; I may be working on something completely different in a month. That’s a big motivation, and what sets us apart as we tackle basic questions that can have a big impact,” Paquit said.
ORNL is managed by UT-Battelle for DOE’s Office of Science, the single largest supporter of basic research in the physical sciences in the United States. DOE’s Office of Science is working to address some of the most pressing challenges of our time. For more information, please visit http://science.energy.gov.