BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VTIMEZONE
TZID:Europe/Vienna
BEGIN:DAYLIGHT
DTSTART:20130331T030000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20121028T020000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260505T043308Z
UID:506956deaad9d@ist.ac.at
DTSTART:20121112T163000
DTEND:20121112T173000
DESCRIPTION:Speaker: Jan Peters\nAbstract: Autonomous robots that can assis
 t humans in situations of daily life have been a long standing\nvision of 
 robotics\, artificial intelligence\, and cognitive sciences. A first step 
 towards this goal is to\ncreate robots that can learn tasks triggered by e
 nvironmental context or higher level instruction.\nHowever\, learning tech
 niques have yet to live up to this promise as only few methods manage to\n
 scale to high-dimensional manipulator or humanoid robots. In this talk\, w
 e investigate a general\nframework suitable for learning motor skills in r
 obotics which is based on the principles behind\nmany analytical robotics 
 approaches. It involves generating a representation of motor skills by\npa
 rameterized motor primitive policies acting as building blocks of movement
  generation\, and a\nlearned task execution module that transforms these m
 ovements into motor commands. We\ndiscuss learning on three different leve
 ls of abstraction\, i.e.\, learning for accurate control is\nneeded to exe
 cute\, learning of motor primitives is needed to acquire simple movements\
 , and\nlearning of the task-dependent "hyperparameters" of these motor pri
 mitives allows learning\ncomplex tasks. We discuss task-appropriate learni
 ng approaches for imitation learning\, model\nlearning and reinforcement l
 earning for robots with many degrees of freedom. Empirical\nevaluations on
  a several robot systems illustrate the effectiveness and applicability to
  learning\ncontrol on an anthropomorphic robot arm
LOCATION:Raiffeisen Lecture Hall\, Central Building\, ISTA
ORGANIZER:mtrappl@ist.ac.at
SUMMARY:Jan Peters: The Institute Colloquium: Towards motor skill learning 
 for robotics
URL:https://talks-calendar.ista.ac.at/events/467
END:VEVENT
END:VCALENDAR