Adjust Font Size: Normal Large X-Large

Return to Events List


Special Seminar: Robust Input-Output Stability: Illustrative Examples and Theoretical Extensions

Ryan Caverly, Graduate Student, Research Assistant, University of Michigan

2:30 PM on 2017-04-04

3-180 Keller Hall


All engineering models used for prediction and controller design are, to some degree, uncertain. Because of this uncertainty, even a well-designed controller can lead to poor closed-loop performance or instability upon implementation if model uncertainty is not explicitly considered. Robust controllers are often designed for worst-case uncertainty, and can provide a guarantee of robust closed-loop stability and/or performance. This talk will focus on the importance of robust control through the illustrative examples of cable-actuated robotic and aeroelastic systems, as well as recent theoretical results that have launched a new branch of robust control. First, the dynamic modeling and control of a flexible cable-driven parallel manipulator (CDPM) will be presented. Novel agile flight simulators have recently been designed using CDPMs, which has enabled unprecedented flight simulation capabilities, but has also led to challenging control problems. Passivity-based control, a robust control architecture, will be introduced and applied to the CDPM, and be shown to significantly outperform competing state-of-the-art control techniques in the presence of model uncertainty. Other applications, such as wind energy-harvesting kiteplanes and flexible aircraft, will also be discussed. Second, extensions and applications of the Large Gain Theorem, which is a little-known stability result that complements the well-known Small Gain Theorem, will be discussed. The Large Gain Theorem can be used to design robust controllers that outperform state-of-the-art techniques, such as H-infinity control, for certain types of model uncertainty. Moreover, robust controllers designed using the Large Gain Theorem can robustly stabilize systems with unstable uncertainty models, which is a feat that no existing robust control technique can match.

Bio:

Ryan James Caverly received his B.Eng. in Mechanical Engineering (Honours) from McGill University,Montreal, QC, Canada, in 2013. Ryan was awarded his M.Sc.Eng. in Aerospace Engineering from the University of Michigan, Ann Arbor, MI, in 2015, where he will soon complete his Ph.D. in Aerospace Engineering. His doctoral research has involved the development of robust controller synthesis methods that employ the little-known Large Gain Theorem. Controllers designed using these new synthesis methods have been found to outperform those designed using more traditional robust controllers in certain situations. Ryan has also worked on collaborative projects with NASA Armstrong and Systems Technology, Inc. involving disturbance observer design for the gustload alleviation of high-altitude long-endurance (HALE) aircraft. Ryan has received multiple best presentation in session awards at national conferences, and is the recipient of Master’s and doctoral NSERC fellowships from the Canadian government. His research explores the intersection of dynamics and control theory, with a focus on robust control and applications to aerospace, mechanical, and marine robotic systems.


Return to Events List