Topic: The Foundation of Intelligent Decision-Making: Model Predictive Control and Optimization Algorithms
Speaker: Liang Wu (Johns Hopkins University, USA)
Abstract
Through model-based prediction, receding-horizon optimization and feedback correction, Model Predictive Control (MPC) incorporates system performance and physical constraints into the decision-making process in a unified manner, serving as an important methodology that bridges systems theory, control theory, numerical optimization and artificial intelligence. This lecture first introduces the basic principles and application scenarios of MPC, elaborating on the connections among prediction models, optimization solvers and closed-loop control. It then draws on the speaker's own research to focus on efficient quadratic programming algorithms, execution-time certification and parallel optimization along the prediction horizon, presenting algorithm design ideas for real-time control. The lecture further explores Koopman-based predictive modeling, differentiable predictive control and learning-based optimization, discussing the fusion of model-driven and data-driven approaches, as well as the connections between online optimization and policy learning. Finally, with a view to robotics and autonomous systems, it discusses how to balance solving efficiency, constraint satisfaction and control performance under limited computational budgets, and outlines development directions in which optimization and learning jointly support real-time, reliable and intelligent decision-making.
Speaker Bio
Liang Wu is currently a postdoctoral researcher at Johns Hopkins University, USA, and will soon join the School of Mathematics at Southeast University. He received his PhD in Computational Sciences and Systems Engineering from the IMT School for Advanced Studies Lucca, Italy, under the supervision of Prof. Alberto Bemporad, a renowned scholar in the field of model predictive control. He then worked as a postdoctoral researcher at the Massachusetts Institute of Technology, USA, under Prof. Richard D. Braatz, a Member of the U.S. National Academy of Engineering. His research has long focused on model predictive control, optimization algorithms and learning algorithms. He has published more than 20 papers in total, including, over the past five years, five first-author papers (two of them full-length papers) in IEEE Transactions on Automatic Control, a premier journal in the control field; two first-author papers in IEEE Control Systems Letters; and one first-author paper at NeurIPS, a premier conference.
Time: Wednesday, October 21, 2026, 15:30–17:10
Venue: Undergraduate Classroom, 4th Floor, Zhengxin Building, Qianwei Campus, Jilin University
Online access (meeting ID): 439-6474-5413
