Basser Seminar Series

Machine Learning and Optimization for Robotics

Speaker: Associate Professor Pieter Abbeel
Department of Electrical Engineering and Computer, UC Berkeley

When: Monday 23 February 2015, 10-11am, *Note different day and time to usual.

Where: The University of Sydney, School of IT Building, SIT Lecture Theatre (Room 123), Level 1

Add seminar to my diary

Abstract

In the first half of this talk I will describe apprenticeship learning, in which robots learn as apprentices from watching human demonstrations. Apprenticeship learning has enabled the most advanced helicopter aerobatics to-date, including maneuvers such as chaos, tic-tocs, and auto-rotation landings which only exceptional expert human pilots can fly, as well as challenging robotic manipulation skills, such as knot-tying and cloth manipulation.

In the second half of this talk I will describe our work on deep reinforcement learning. Reinforcement learning considers the problem of learning control policies through a directed process of trial and error. It has already had a moderate number of success stories in robotics. However, substantial domain expertise was required to craft good control policy architectures with a relatively small number of free parameters. Rather than handcrafting such architectures deep reinforcement learning carries the premise to directly optimize a deep, layered mapping with thousands, or even millions, of parameters from recent sensory signals to actions. I will describe our guided policy search approach to deep RL, which has already enabled learning a variety of manipulation primitives, such as screwing caps onto bottles, stacking lego-blocks, placing a tight-fitting ring around a pole, and some simple assembly tasks. I will also describe current directions, and preliminary results on learning to walk and learning to play Atari games.

Speaker's biography

Pieter Abbeel received his Ph.D. degree in Computer Science from Stanford University in 2008 and is currently on the faculty at UC Berkeley in the Department of Electrical Engineering and Computer Sciences. He has won various awards, including best paper awards at ICML and ICRA, the Sloan Fellowship, the Air Force Office of Scientific Research Young Investigator Program (AFOSR-YIP) award, the Office of Naval Research Young Investigator Program (ONR-YIP) award, the DARPA Young Faculty Award (DARPA-YFA), the MIT TR35, the IEEE Robotics and Automation Society (RAS) Early Career Award, and the Dick Volz Best U.S. Ph.D. Thesis in Robotics and Automation Award.