Convex Relaxations in Nonlinear System Identification, Analysis, and Control
Dr. Ian Manchester (University of Sydney)
SYSTEMS AND CONTROL SERIESDATE: 2013-11-15
TIME: 11:00:00 - 12:00:00
LOCATION: RSISE Seminar Room, ground floor, building 115, cnr. North and Daley Roads, ANU
CONTACT: JavaScript must be enabled to display this email address.
ABSTRACT:
Computational aspects of many important problems for linear systems are by now quite well understood in terms of solutions to Riccati equations or LMI's. For nonlinear systems the situation is quite different, and most optimisation problems of interest are intractable in their natural form. Convex relaxation is, roughly speaking, the process of finding a "nearby" convex problem and solving that instead. Obviously it is important to understand in what sense it is "nearby", and how solutions of the two problems are related.
This talk will present a range of recent results on identification of stable nonlinear state space models, prediction of limit cycles, and stabilizing control design. The principle relaxation techniques are sum-of-squares programming, the S-Procedure (aka Lagrangian relaxation), dissipation inequalities, and contraction analysis. The methods will be illustrated with applications to control of bipedal walking robots, computational neuroscience, and clinical neurology.
BIO:
Ian R. Manchester received BE and PhD degrees in electrical engineering from the University of New South Wales in 2002 and 2006, respectively. From 2006-2009 he was a post-doctoral researcher at Umea University, Sweden, and from 2009-2012 he was a Research scientist at the Massachusetts Institute of Technology, in the Robot Locomotion Laboratory and the Laboratory for Information and Decision Systems. Since 2012 he has been a Senior Lecturer the University of Sydney, Australia.
Currently, his research is focused on theoretical and algorithmic methods for nonlinear systems. In particular: control design for walking robots, contraction analysis of oscillations, identification of stable nonlinear models, and input design for system identification. His recent theoretical work is applied to and motivated by applications in clinical neurology, computational neuroscience, mining automation, and commercial aviation.
