BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:An Optimization-based Approach to Safe and Efficient Learning-base
 d Control - Melanie Zeilinger\, ETH Zürich
DTSTART:20231012T131500Z
DTEND:20231012T141500Z
UID:TALK207079@talks.cam.ac.uk
CONTACT:Fulvio Forni
DESCRIPTION:Numerous demonstrations have highlighted the potential of lear
 ning-based control paradigms over the last years. A number of both practic
 al and theoretical challenges arising from the use of learning in closed-l
 oop control systems\, however\, are still limiting the widespread success 
 of these promising techniques in practice. In this talk\, I will focus on 
 the questions of data-efficiency and safety. In order to provide a scalabl
 e and flexible approach for complex systems\, the techniques will build on
  optimization-based control. The first part of the talk will discuss techn
 iques for learning objective and dynamics models\, which are both real-tim
 e capable and allow for fast online adaptation despite limited initial dat
 a and information for a particular task at hand. In the second part\, I wi
 ll present the concept of a predictive safety filter to augment any learni
 ng-based controller with safety guarantees and will outline connections to
  techniques based on control barrier functions. The results will be highli
 ghted using examples from robotics.\n\nThe seminar will be held in the JDB
  Seminar Room\, Department of Engineering\, and online (zoom): https://new
 nham.zoom.us/j/92544958528?pwd=YS9PcGRnbXBOcStBdStNb3E0SHN1UT09
LOCATION:Department of Engineering / Online (Zoom)
END:VEVENT
END:VCALENDAR