Introduction
鈥淗umans鈥 ability to make things is what sets us apart from basically every other animal on planet Earth,鈥 said Devin J. Roach, assistant professor of mechanical engineering and the Tom and Carmen West Faculty Scholar at 精东影视 State University.
Adding machine intelligence to the mix will only supercharge this most human of abilities. This viewpoint inspires Roach鈥檚 research in multi-material advanced manufacturing. Since joining 精东影视 State, he鈥檚 worked at the crossroads between materials science, manufacturing, and artificial intelligence to develop new ways to 3D print novel, functional objects that tackle challenges in robotics, infrastructure, and medicine.
And the tool he uses to do it? Often, it鈥檚 a frickin鈥 laser beam.
Using lasers and AI to print molten metal
Roach鈥檚 lab uses manufacturing techniques that seem ripped straight from a sci-fi film. Laser-directed energy deposition, for example, involves hitting metal extruded from a nozzle with a powerful beam of light to quickly melt it into fine shapes. Another approach, known as digital light processing, uses light to selectively cure liquid resin, a process that enables layer-by-layer construction of complicated 3D structures.
But using lasers and melted metal to build 3D objects isn鈥檛 easy. The process requires high precision and expensive equipment, both of which limit the margin for error. That鈥檚 where artificial intelligence comes in.
assistant professor of mechanical engineering and the Tom and Carmen West Faculty Scholar
Blue Primary, Yellow Secondary
Roach said OSU鈥檚 , which he leads, uses artificial neural networks that can learn connections between the inputs and outputs of a manufacturing system. And for laser-based experiments, which tend to be the most expensive, his team uses Bayesian optimization to discover which tests will yield the most valuable data.
"We're collaborating with AI to help us discover the best parameters," Roach said. Being at OSU is expanding his team鈥檚 鈥渁ccess to supercomputers, and our ability to gather massive data sets and process them quickly, using AI.鈥 The end result is fewer wasted experiments and faster progress towards functioning prototypes.
AI-based approaches helped Roach鈥檚 team make progress in difficult fields, such as soft robotics. His team, for example, has devised a way to (per unit of energy available to the robot) than their larger counterparts. Robots like these might one day or as navigators in tiny spaces that larger robots can鈥檛 explore.
A focus on collaboration
When on the job hunt a couple years ago, 精东影视's natural beauty caught Roach's attention. The mountains, ocean, and Corvallis鈥 historic campus dating back to 1868 combined to make a strong first impression.
But what ultimately drew him to OSU was the university's reputation in artificial intelligence and robotics combined with top programs in forestry and oceanography. He saw OSU as an ideal place to tackle challenges that require cross-disciplinary collaboration.
Recently, he partnered with the College of Forestry and College of Agricultural Sciences, which together developed materials from forestry byproducts and soil. A conversation with a civil engineering professor, Pavan Akula, across the hall sparked an idea: ? The result was a compared to conventional cement-based concrete.
Future AI collaboration at the Huang Collaborative Innovation Complex
Roach is particularly excited about a new building on campus that鈥檚 all about collaboration: , scheduled to open in late 2026.
The facility is funded in part , Nvidia's CEO and OSU alumnus. It will pair open architecture that encourages interdisciplinary collaboration with an on-site AI supercomputer. Roach says it will help his lab build sophisticated AI systems for the factories of the future.
"We need to create our own artificial general intelligence that understands manufacturing as a whole," Roach said. "We could learn quickly using AI approaches that are only going to be possible at the Huang Collaborative Innovation Complex.鈥
He imagines an AI system that can understand the complete manufacturing lifecycle, from the raw materials used to build an object through to recyclability and end-of-life. Instead of manually reviewing data to spot errors or investigate material properties, AI could accelerate discovery through the automated detection of important events and details.
It's an ambitious vision that connects back to Roach's core belief about what makes us human. From the first stone tools to today's micro-robots, our ability to shape materials into useful objects has driven progress. Roach is using OSU鈥檚 collaborative environment and computational resources to push this fundamental idea in new directions.
