The workshop will be held on May 6th, 2013, in the Kongresszentrum Karlsruhe (Stadthalle, Hans-Thoma-Saal) as part of the 2013 IEEE International Conference on Robotics and Automation.
From Machine Learning to learning in real-world autonomous systems
Machine Learning is playing an increasingly important role in the field of robotics. The list of success stories includes its application on map learning, system identification, motor skill learning, perceptual learning and others. However, there are many open problems when systems are to autonomously learn complex integrated tasks, e.g., preparing a meal or assembling a shelf, even via imitation learning. Learning in challenging robotic domains often involves finding the right decomposition and representation of problems, coping with very high-dimensional, hybrid or relational spaces, and dealing jointly with uncertainty, geometry and logic. These challenges are particular to robotics and the structure of robotic environment and therefore raise genuine challenges for Machine Learning.
The aim of this is to bring together researchers to discuss the current state-of-the-art of learning in robotic systems, including its limitations and future directions. The topics of interest include:
- Pushing Machine Learning towards learning in autonomous robots – learning to enable autonomous behavior in natural environments
- Autonomous exploration and learning
- Learning for object manipulation
- Planning and control with learned uncertain models
- Reinforcement Learning and inverse RL in robotics
8:30 — 8:45 Welcome
8:45 — 9:30 Invited Talk: Mrinal Kalakrishnan (Stefan Schaal’s lab, University of Southern California) Learning Objective Functions for Autonomous Robotic Manipulation
9:30 — 10:00 Contributed Paper: Henrik Kretzschmar, Markus Kuderer, Wolfram Burgard: Inferring Navigation Policies for Mobile Robots from Demonstrations)
10:00 — 10:30 Coffee Break
10:30 — 11:15 Invited Talk (Nicholas Roy, Massachusetts Institute of Technology, USA)
11:15 — 11:45 Contributed Paper: Rudolph Triebel, Hugo Grimmett, Ingmar Posner: Confidence Boosting: Improving the Introspectiveness of a Boosted Classifier for Efficient Learning
11:45 — 12:30 Contributed Paper: Peter Englert, Alexandros Paraschos, Jan Peters, Marc Deisenroth: Addressing the Correspondence Problem by Model-based Imitation Learning
12:30 — 14:00 Lunch Break
14:00 — 14:45 Invited Talk: Siddhartha Srinivasa (Carnegie Mellon University) Weakly Supervised Robotic Object Discovery (see also paper and video)
14:45 — 15:15 Contributed Paper: Thomas Witzig, Rainer Jaekel, Dejan Pangercic, Sarah Osentoski, Marius Zoellner, Ruediger Dillman: Context Aware Shared Autonomy for Robotic Manipulation Tasks
15:15 — 15:35 Poster-Spotlights (10×2 minutes)
15:35 — 16:00 Coffee Break & Poster Session
16:00 — 16:45 Poster Session
16:45 — 17:30 Invited Talk (Sethu Vijayakumar, University of Edinburgh)
17:30 — 18:00 Contributed Paper: Renaud Detry, Martin Hjelm, Carl Henrik Ek, Danica Kragic: Generalizing Task Parameters Through Modularization
18:00 — 18:15 Closing Remarks
Call for Paper and Extended Abstracts
We invite the submission of papers and extended abstracts. Upon acceptance, if submitted by Feb. 25th contributions will be included in the official ICRA workshop proceedings; late-breaking papers can be submitted until 15. April. All accepted contributions will be published on the workshop webpage. Authors of accepted paper will present their work during the interactive Poster sessions, preceded by a two-minutes spotlight presentation. We will employ light-touch reviewing. Everybody submitting a paper will be expected to also review a paper. Criteria are originality of ideas and relevance to the topics of Autonomous Learning.
Deadlines for inclusion in the ICRA workshop proceedings
Submission: 25. February 2013, Midnight
Notification: 6th March 2013
Final version: 11th March 2013
Deadlines for late-braking papers
Submission: 15. April 2013, Midnight
Notification: 24. April 2013
Final version: 29. April 2013
Instructions for presenters
All accepted contributions to the AL workshop can present their work as a poster. For this, please prepare a poster in Portrait (not Landscape) format, size A1 or A0. If you like you can also use your laptop to present videos.
Those contributions not selected for oral presentations will have a 2 minutes spotlight presentation before the poster session to attract people to their work. For this please prepare a SINGLE pdf slide and send us the slide by May 2nd. Also, prepare well to not exceed the 2 minutes during the spotlight.
[wp-publications bib=”alw_papers.bib” all=true]