Joint Colloquium by Electrical & Computer Engineering and Institute of Advanced Systems Engineering
October 21, 2022 12:30-1:30 P.M.
Virtual via Webex
https://uconn-cmr.webex.com/uconn-cmr/j.php?MTID=m9ecacb8417524edc017aed6e45bb2384
Adaptive Autonomous Systems: A Control Theory Perspective
Dr. Sayan Basu Roy
Assistant Professor of Electrical and Computer Engineering
Indraprastha Institute of Information Technology, Delhi, India
Abstract: Adaptive Control is an elegant nonlinear control technique which can deal with parametric uncertainty in real-time while ensuring Lyapunov stability in the extended state-space of tracking error and parameter estimation error. It is well-known that transient performance and efficient parameter learning of closed-loop adaptive autonomous systems are heavily dependent on an analytical condition, called persistence of excitation (PE). The PE condition is restrictive in nature due to its reliance on the future evolution of the corresponding signal and it is difficult to satisfy and/or verify the condition in real-world applications. This talk will focus on a new class of adaptive control algorithms, which have been developed over the past few years to relax the PE condition without sacrificing the closed-loop performance. A new condition,called Initial Excitation (IE) is defined and it is shown to be sufficient for exponential stability in the extended state-space using a filter-based adaptive control formulation. The talk will also cover some of the recent extensions of the IE condition in the context of adaptive LQR, adaptive extremum seeking and distributed adaptive control in brief.
Bio: Sayan is an Assistant Professor at ECE Department of Indraprastha Institute of Information technology (IIIT) Delhi, India since Sept. 2018. Prior to that, he obtained a PhD degree from the Control and Automation group of Electrical Engineering at Indian Institute of Technology (IIT) Delhi (2015-2018). He has a bachelor’s degree in Electronics and Tele-communication from Jadavpur University (2010-2014). He is primarily an applied mathematician working on the broad domain of nonlinear control theory and dynamical systems. He deals with various aspects of adaptive and learning controller design like stability, transient response, optimality and robustness especially for safety-critical dynamical systems. He is currently spending time as a visiting researcher at the Mechanical Engineering Department of Massachusetts Institute of Technology (MIT), USA with a focus on accelerated learning algorithms for online identification and adaptive control problems.
For more information, contact: Brandy Ciraldo at brandy.ciraldo@uconn.edu