Building 53, Room 4025
Speaker: Dr Pritesh Patel, University of Southampton
Title: Memory Augmented Adaptive Identification and Control of Switched Linear Systems
Abstract:
I am a Research Fellow at Cyber-Physical System Group, University of Southampton, working with Dr. Andrea Lechini Visintini and Professor Mathew Turner. I earned my Bachelor's in Electronics and Telecommunication from the National Institute of Technology (NIT), Raipur, India and my Masters under the supervision of Prof. Janardhanan Sivaramakrishnan and PhD under the supervision of Prof. Shubhendu Bhasin in Control and Automation, Electrical Engineering, Indian Institute of Technology (IIT) Delhi, India. My research interests are adaptive and learning control, switched dynamical systems and adaptive fault tolerant control with applications in aerospace and robotics.
Abstract: Achieving convergence of parameter estimates to their true values is a crucial aspect of adaptive system design, enhancing performance and robustness. However, this convergence typically relies on the persistence of excitation (PE) condition, which is difficult to satisfy in practice. The problem of parameter convergence becomes even more challenging when dealing with switching parameters, as traditional adaptive methods struggle to ensure convergence without PE.
In my talk, I will discuss how memory augmentation and intermittent initial excitation (IIE) facilitate parameter learning even during the inactive phases of a subsystem. These techniques enable parameter convergence for switched linear systems, ultimately ensuring global exponential stability (GES) of the overall error, which includes tracking and parameter estimation errors of all subsystems.