Trustworthy Agentic AI for Multidisciplinary Design Optimization
Abstract: Agentic artificial intelligence (AI) has emerged as a powerful tool for accelerating scientific discovery and the design of high-performance engineered systems. However, a key question remains: How can we develop AI agents that produce trustworthy results and can be reliably incorporated into our daily research workflows? In this talk, I will introduce our approach to building a trustworthy agentic AI framework for multidisciplinary design optimization (MDO) with a focus on the measures needed to ensure reliable and reproducible results.
The first part of the talk will present our scalable MDO framework, which integrates high-fidelity solvers for aerodynamics, structures and heat transfer. We develop efficient discrete adjoint algorithms to enable gradient-based optimization with hundreds of design variables and constraints, making large-scale MDO practical. A range of MDO applications, including aircraft, propellers and thermal systems, will be presented to demonstrate the versatility and robustness of the framework.
The second part of the talk will focus on building AI agents to enable end-to-end MDO workflows, including geometry manipulation, mesh generation, simulation, optimization and post-processing. I will discuss the guardrails we use to improve trustworthiness, including domain-specific agents and skills, curated input parameters, robust review and correction mechanisms, and fully auditable and reproducible workflows. These measures allow the agentic MDO workflow to achieve deterministic behavior and a high success rate.
Finally, I will demonstrate the agentic MDO workflow through several examples, including wing aerostructural optimization, propeller aeroacoustic optimization and unmanned aerial vehicle trajectory optimization. This paradigm has the potential to broaden access to advanced MDO capabilities and significantly accelerate engineering design.
Bio: Ping He is an assistant professor of aerospace engineering at Iowa State University (ISU). Prior to joining ISU, he conducted postdoctoral research at the University of Michigan and at North Carolina State University. He earned his Ph.D. from the Chinese Academy of Sciences. He recently received the National Science Foundation CAREER Award.
Series: See all upcoming talks at https://www.anl.gov/mcs/lans-seminars.