Designing new aircraft has traditionally required engineers using software to generate thousands of intermediate potential configurations and run extensive computer simulations to determine which designs will perform reliably and effectively under real-world conditions. This approach is essential for safety and performance, but it also demands significant time, computing resources and engineering effort.
Dr. Xiaosong Du, assistant professor of mechanical and aerospace engineering, is pioneering a more efficient path forward by incorporating generative artificial intelligence (AI) into the earliest stages of the design process. His goal is to dramatically filter out the number of unrealistic configurations and streamline development workflows for next-generation aircraft and drones.
“Many people think of generative AI as a tool for writing text or creating images, but its potential goes far beyond that,” says Du. “By developing AI that can generate aircraft or drone designs already built to meet requirements, we can skip thousands of failed tests and streamline processes.”
With support from the National Science Foundation, Du is developing a system that trains generative AI models using flight-related data — including aerodynamic performance, structural limits and operational constraints — and then pairs those models with predictive tools that can rapidly evaluate whether newly generated designs satisfy strict engineering conditions.
The system creates a rapid feedback loop: the generative AI proposes new configurations, predictive models evaluate them instantly and the AI refines its approach based on that response. Over time, this enables the AI to produce only viable design candidates right from the start, eliminating much of the iterative testing that currently dominates aerospace design.
He says this iterative learning process will help the AI “learn to produce viable designs from the start and eliminate much of the trial-and-error that is currently common in his field.”
Beyond saving time, Du believes this approach could significantly enhance innovation across the aerospace sector. “As this generative AI tool takes off in the coming years, it could reshape how aerospace engineers design next-generation flight systems and help us see even faster progress in the field,” Du says. By advancing generative AI specifically for engineering applications, Dr. Du’s research positions Missouri S&T at the forefront of intelligent aerospace design — enabling faster, smarter and more efficient development of future flight technologies.