Upcoming Event: Center for Autonomy Seminar

Why do we need "control" in control oriented learning?

Mario Sznaier, Northeastern University, Boston

Abstract

Despite recent advances in Machine Learning (ML), the goal of designing control systems capable of fully exploiting the potential of these methods to learn from the environment and safely achieve complex specifications remains elusive: we have been told that "fully autonomous vehicles are less than 5 years away"...since 2015.  Modern ML methods can leverage large amounts of data to learn powerful predictive models, but such models are not designed to operate in a closed-loop environment. Recent results on reinforcement learning offer a tantalizing view of the potential of a rapprochement between control and learning, but so far proofs of performance are restricted to limited cases (e.g. finite horizon LQR/LQG). Thus, in most cases learning elements are used as black boxes within the loop, with less than completely understood properties. Further progress hinges on the development of a principled understanding of the properties and limitations of ML algorithms when used in a control systems context. 

 In the first  portion of the talk we argue that when the goal is to learn models that can be used for control,  loss functions based on open loop metrics are not adequate. Rather, what is needed are metrics that capture closed-loop distances, such as the gap metric introduced in the 1990's. We will conclude this portion of the  talk by presenting some recent results on gap-metric based learning and discuss generalizations.

 In the second part of the talk we will address the problem of designing non-linear controller directly from data, bypassing the identification steps.  We will start by  presenting  some simple examples where commonly used ML techniques (e.g. Deep Learning, Reinforcement Learning) will provably fail to find a stabilizing controller and discuss the implications of these results for the type of architectures needed for control.   We will conclude the talk by presenting an overview of  data-driven safe control of non-linear systems and point out to open problems.

Biography

Mario Sznaier is currently the Dennis Picard Chaired Professor at the Electrical and Computer Engineering Department, Northeastern University, Boston. Prior to joining Northeastern University, Dr. Sznaier was a Professor of Electrical Engineering at the Pennsylvania State University and also held visiting positions at the California Institute of Technology. His research interest include data driven and learning enabled control,  robust identification and control of hybrid systems, and dynamical vision.  Dr. Sznaier is currently serving as Editor in Chief of the section on AI and Control of the journal Frontiers in Control Engineering and Vice-Chair of  IFAC’s Design Methods Technical Activities. Additional recent service include General Chair of the 2024 SysID, Chair of IFAC's TC 2.5 (Robust control), Associate Editor for the Journal Automatica (2005-2021), and Program Chair of the 2017 CDC.  He is a distinguished member of the IEEE Control Systems Society  and a Fellow of the IEEE for his contributions to robust control, identification and dynamic vision.