Upcoming Event: Center for Autonomy Seminar

Data-enabled learning for personalized predictive control of nonlinear systems

Marzia Cescon, David C. Zimmerman Assistant Professor of Mechanical and Aerospace Engineering, University of Houston

Abstract

The design of control laws that directly exploit experimental data collected from the process to be controlled has attracted significant attention in the control community for several decades. Recently, model-free, data-driven techniques have emerged to synthesize and tune controllers using large amounts of data. While appealing, these methods face practical limitations due to the quantity and quality of the available data.

In this presentation, I will explore an alternative approach to data-based control design for nonlinear systems with unknown dynamics. First, I will introduce the use of neural multi-step-ahead predictors, identified from data, within a receding-horizon optimal control framework. Next, I will discuss the auto-calibration of the resulting controller as a black-box global optimization problem that relies on a digital twin of the process. Finally, I will present results demonstrating the application of this framework to individual-specific automated insulin delivery for glucose regulation in people with type 1 diabetes.

Biography

Marzia Cescon is the David C. Zimmerman Assistant Professor of Mechanical and Aerospace Engineering at the University of Houston (UH). At UH she is also founder and director of the Advanced Learning, Artificial Intelligence and Control laboratory, a multidisciplinary effort for learning-based decision making and control of complex and potentially unknown dynamical systems.  Dr. Cescon earned a bachelor’s degree in information engineering and a master’s degree in control systems engineering from the University of Padua, Italy, and received a Ph.D. in automatic control from Lund University, Sweden. She has held several research positions, including at the University of California at Santa Barbara, the University of Melbourne, and Harvard University. Dr. Cescon is the recipient of the Cornelis Drebbel Faculty Fellowship from TU Delft (2023), the NSF Career Award (2024), the Cullen College of Engineering Early Inventor Award (2025), and the ASME Rising Star in Mechanical Engineering Award (2025).