11/11/2026 Bruce Adams
Yingying Li received a 2026 National Artificial Intelligence Research Resource (NAIRR) Pilot Award for her project, “Stable and Scalable Contextual Reinforcement Learning for Nonstationary and Networked Systems.” The award supports the project through the NAIRR Pilot. Li's research areas include reinforcement learning, learning-enabled control and multi-agent/network decision-making.
Written by Bruce Adams
We have achieved a lot in the simulator, in the lab, and even in some real-world experiments. But when we apply AI to real-world scenarios, we still see failure cases, such as robots or autonomous vehicles not performing as well as we wished on a snowy day or a rainy day.
—Yingying Li, Assistant Professor, Industrial and Enterprise Systems Engineering
Yingying Li received a 2026 National Artificial Intelligence Research Resource (NAIRR) Pilot Award for her project, “Stable and Scalable Contextual Reinforcement Learning for Nonstationary and Networked Systems.” The award supports the project through the NAIRR Pilot. Li's research areas include reinforcement learning, learning-enabled control and multi-agent/network decision-making.
“In open-ended and highly complex environments, there are lots of uncertainties that are unexpected in the training phase. AI agent training does not capture how agents should handle those uncertainties. This degrades performance and, most importantly, causes safety and reliability issues. That's a challenge we are still facing,” said Li, a professor of industrial and enterprise systems engineering in the University of Illinois Urbana-Champaign's Grainger College of Engineering.
The project aims to develop more reliable reinforcement learning algorithms that can adapt to changing real-world conditions while maintaining stability and safety guarantees. By combining reinforcement learning with techniques from control theory, Li's team seeks to enable AI agents to adjust their decision-making strategies as environments evolve, rather than relying on a fixed strategy learned during training. researchers will evaluate their methods in large-scale simulations involving robotics, power systems, and multi-agent networks, where unexpected changes and communication disruptions can significantly affect system performance.
"My group has been working for several years on topics on reliable AI-enabled autonomous systems,” she said.
Li’s research team includes Ph.D. students Haonan Xu, Ziyao Guo, Ghazal Hassani, Shengkang Gao, Yulun Wu and Haori Lu (who are co-advised by Yaoyao Liu) and undergraduate student Alexandra Li.
The researchers will use GPU computing resources provided through the NAIRR Pilot on National Center for Supercomputing Applications' Delta and DeltaAI systems to conduct large-scale experiments evaluating the stability and reliability of reinforcement learning algorithms in changing environments.
The National Artificial Intelligence Research Resource Pilot (NAIRR) is a project creating a shared national research infrastructure to bridge research and training gaps by connecting United States researchers and educators with computation, data, software, models, training, and educational materials to advance research, discovery, and innovation.
Grainger Engineering Affiliations
Yingying Li is an Illinois Grainger Engineering assistant professor of industrial and enterprise systems engineering, an affiliate professor in the School of Information Sciences, and is affiliated with Electronic and Computer Engineering, Mechanical Science & Engineering and Coordinated Science Lab.
Yaoyao Liu is an assistant professor of information sciences and is affiliated with the Siebel School of Computing and Data Science, Electronic and Computer Engineering, Coordinated Science Lab and National Center for Supercomputing Applications.