Frontiers in Networked Control

[Instructor] [Announcements] [Schedule] [Description]

Instructor

Michael Rotkowitz
Email: FIRSTNAME.rotkowitz@ee.kth.se
Phone: 08-790-8778
Office: Osquldas väg 10, Level 6, Room A:613
Homepage: http://www.ee.kth.se/~mcrotk
Back to Top

Announcements

  • [2006.05.27] Speakers, scribes, dates, and topics have been updated. Please inform me of any problems.
  • [2006.05.27] The style file and template for the lecture notes may now be downloaded here.
  • [2006.02.16] The first class will meet on March 1, at the room and time indicated below. If you're considering taking the course, but can't make the first one, please send me an email.
Back to Top

Schedule


Back to Top

Description

Conventional controls analysis assumes that the controllers to be designed all have access to the same measurements. With the advent of complex systems, decentralized control has become increasingly important, where one has multiple controllers each with access to different information. Examples of such systems include flocks of aerial vehicles, autonomous automobiles on the freeway, the internet, the power distribution grid, spacecraft moving in formation, and paper machining. Conventional methods of analysis and synthesis break down for a decentralized system.

This course will survey different methods of attacking the general difficulty of decentralized control problems. Each week, we will review one paper, or a set of related papers, from the recent literature. Each student will present the lecture once, and will be able to choose a topic of particular interest to them. There will be a list of suggested topics / papers, but others may be selected if they fit with the theme of the course and represent a fundamentally different approach from the other topics.

The course will be available for 2 or 4 credits. The requirements for the first two credits are presenting one lecture (primarily on the whiteboard), writing up (in LaTeX) the lecture notes for someone else’s lecture, and regular attendance. Two additional credits may then be earned for completing a course project.

Please forward this to anyone who may be interested who I may have missed. While the focus will be on control under imperfect information, several of the topics apply to difficult optimization problems much more generally, and so it could be of interest to a broader audience. Of course, let me know if you have any questions.
Back to Top

2006.08.13