BEGIN:VCALENDAR VERSION:2.0 PRODID:researchseminars.org CALSCALE:GREGORIAN X-WR-CALNAME:researchseminars.org BEGIN:VEVENT SUMMARY:Jan Peters (Technische Universitaet Darmstadt) DTSTART:20210423T130000Z DTEND:20210423T140000Z DTSTAMP:20260423T003251Z UID:MPML/33 DESCRIPTION:Title: Ro bot Learning - Quo Vadis?\nby Jan Peters (Technische Universitaet Darm stadt) as part of Mathematics\, Physics and Machine Learning (IST\, Lisbon )\n\n\nAbstract\nAutonomous robots that can assist humans in situations of daily life have been a long standing vision of robotics\, artificial inte lligence\, and cognitive sciences. A first step towards this goal is to cr eate robots that can learn tasks triggered by environmental context or hig her level instruction. However\, learning techniques have yet to live up t o this promise as only few methods manage to scale to high-dimensional man ipulator or humanoid robots. In this talk\, we investigate a general frame work suitable for learning motor skills in robotics which is based on the principles behind many analytical robotics approaches. It involves generat ing a representation of motor skills by parameterized motor primitive poli cies acting as building blocks of movement generation\, and a learned task module that transforms these movements into motor commands. We discuss le arning on three different levels of abstraction\, i.e.\, learning for accu rate control is needed to execute\, learning of motor primitives is needed to acquire simple movements\, and learning of the task-dependent „hyper parameters“ of these motor primitives allows learning complex tasks. We discuss task-appropriate learning approaches for imitation learning\, mode l learning and reinforcement learning for robots with many degrees of free dom. Empirical evaluations on a several robot systems illustrate the effec tiveness and applicability to learning control on an anthropomorphic robot arm. These robot motor skills range from toy examples (e.g.\, paddling a ball\, ball-in-a-cup\, juggling) to playing robot table tennis against a h uman being and manipulation of various objects.\n LOCATION:https://researchseminars.org/talk/MPML/33/ END:VEVENT END:VCALENDAR