Jan Peters

Research Group Leader

Empirical Inference

Max-Planck-Ring 4
Tübingen
Germany

+49 7071 601 551

+49 7071 601 552

http://www.ias.informatik.tu-dar...

Jan Peters is a full professor (W3) for Intelligent Autonomous Systems at the Computer Science Department of the Technische Universitaet Darmstadt and at the same time a senior research scientist and group leader at the Max-Planck Institute for Intelligent Systems, where he heads the interdepartmental Robot Learning Group. Jan Peters has received the Dick Volz Best 2007 US PhD Thesis Runner-Up Award, the 2012 Robotics: Science & Systems - Early Career Spotlight, the 2013 INNS Young Investigator Award, and the IEEE Robotics & Automation Society‘s 2013 Early Career Award. In 2015, he was awarded an ERC Starting Grant.

Jan Peters has studied Computer Science, Electrical, Mechanical and Control Engineering at TU Munich and FernUni Hagen in Germany, at the National University of Singapore (NUS) and the University of Southern California (USC), and been an visiting researcher at the ATR Telecomunications Research Center in Japan. He has received four Master‘s degrees in these disciplines as well as a Computer Science Ph.D. from USC.

If you are interested in working with me, please either apply through http://www.ausy.tu-darmstadt.de/Main/OpenPositions or send an official application to Ms. Mara Cascianelli <mara.cascianelli@tuebingen.mpg.de>. PLEASE REFRAIN FROM SENDING ME YOUR APPLICATION BY EMAIL! You will not get a reply as they will get lost in my aggressive spam filter.

Robot Learning Reinforcement Learning Machine Learning Robotics Biomimetic Systems

I am interested in many topics. Among them are

For more information see my private homepage and on my group's homepage.

Jan Peters is a full professor at Darmstadt University of Technology (TU Darmstadt) heading the intelligent systems group while still being affliated with the MPI heading the robot learning lab.

Jan Peters joined the Max-Planck Institute for Biological Cybernetics in April 2007 as a research scientist. Before joining MPI, Jan Peters studied Electrical Engineering, Computer Science and Mechanical Engineering. He graduated from the University of Hagen in 2000 with a Diplom-Informatiker (German M.Sc. in Computer Science) and from Munich University of Technology in 2001 with a Diplom-Ingenieur in Electrical Engineering (German M.Sc. in Electrical Engineering). In 2000-2001, he spent two semesters as visiting student at National University of Singapore. In 2002, he completed a M.Sc. in Computer Science and, in 2005, a M.S. in Mechanical Engineering both from USC. He joined USC's Computer Science Ph.D. program and the CLMC Lab in Fall 2001. Jan Peters has been a visiting research student at the Department of Robotics at the German Aerospace Research Center in Germany, at Siemens Advanced Engineering in Singapore and at the Department of Humanoid Robotics and Computational Neuroscience at the Advanded Telecommunication Research (ATR) Center in Japan. Jan Peters graduated from University of Southern California with a Ph.D. in Computer Science in March 2007. He remains affiliated with the CLMC Lab as an adjunct researcher.

More information on Jan Peters can be found on his private homepage.

Reinforcement Learning

Reinforcement learning ranks among the biggest challenges for machine learning. Just controlling a known dynamical system is hard on its own -- interacting with an unknown system poses even harder decision problems, such as the infamous exploration-exploitation tradeoff. Most research in this area is still conï¬ned to theoretical...

Jan Peters Carl Edward Rasmussen Marc P. Deisenroth Philipp Hennig Oliver Kroemer Jens Kober Yevgeny Seldin Abdeslam Boularias

274 results (View BibTeX file of all listed publications)

2010

Reinforcement Learning by Relative Entropy Policy Search

Peters, J., Mülling, K., Altun, Y.

30th International Workshop on Bayesian Inference and Maximum Entropy Methods in Science and Engineering (MaxEnt 2010), 30, pages: 69, July 2010 (poster)

Abstract

Policy search is a successful approach to reinforcement learning. However, policy improvements often result in the loss of information. Hence, it has been marred by premature convergence and implausible solutions. As first suggested in the context of covariant policy gradients, many of these problems may be addressed by constraining the information loss. In this book chapter, we continue this path of reasoning and suggest the Relative Entropy Policy Search (REPS) method. The resulting method differs significantly from previous policy gradient approaches and yields an exact update step. It works well on typical reinforcement learning benchmark problems. We will also present a real-world applications where a robot employs REPS to learn how to return balls in a game of table tennis.

PDF [BibTex]

2010

Peters, J., Mülling, K., Altun, Y. Reinforcement Learning by Relative Entropy Policy Search 30th International Workshop on Bayesian Inference and Maximum Entropy Methods in Science and Engineering (MaxEnt 2010), 30, pages: 69, July 2010 (poster)

PDF [BibTex]

Grasping with Vision Descriptors and Motor Primitives

Kroemer, O., Detry, R., Piater, J., Peters, J.

In Proceedings of the 7th International Conference on Informatics in Control, Automation and Robotics (ICINCO 2010), pages: 47-54, (Editors: Filipe, J. , J. Andrade-Cetto, J.-L. Ferrier), SciTePress , Lisboa, Portugal, 7th International Conference on Informatics in Control, Automation and Robotics (ICINCO), June 2010 (inproceedings)

Abstract

Grasping is one of the most important abilities needed for future service robots. Given the task of picking up an object from betweem clutter, traditional robotics approaches would determine a suitable grasping point and then use a movement planner to reach the goal. The planner would require precise and accurate information about the environment and long computation times, both of which may not always be available. Therefore, methods for executing grasps are required, which perform well with information gathered from only standard stereo vision, and make only a few necessary assumptions about the task environment. We propose techniques that reactively modify the robots learned motor primitives based on information derived from Early Cognitive Vision descriptors. The proposed techniques employ non-parametric potential fields centered on the Early Cognitive Vision descriptors to allow for curving hand trajectories around objects, and finger motions that adapt to the objects local geometry. The methods were tested on a real robot and found to allow for easier imitation learning of human movements and give a considerable improvement to the robots performance in grasping tasks.

PDF Web Project Page [BibTex]

Kroemer, O., Detry, R., Piater, J., Peters, J. Grasping with Vision Descriptors and Motor Primitives In Proceedings of the 7th International Conference on Informatics in Control, Automation and Robotics (ICINCO 2010), pages: 47-54, (Editors: Filipe, J. , J. Andrade-Cetto, J.-L. Ferrier), SciTePress , Lisboa, Portugal, 7th International Conference on Informatics in Control, Automation and Robotics (ICINCO), June 2010 (inproceedings)

PDF Web Project Page [BibTex]

BCI and robotics framework for stroke rehabilitation

Gomez Rodriguez, M., Grosse-Wentrup, M., Hill, J., Peters, J., Schölkopf, B., Gharabaghi, A.

4th International BCI Meeting, June 2010 (talk)

PDF Web [BibTex]

Gomez Rodriguez, M., Grosse-Wentrup, M., Hill, J., Peters, J., Schölkopf, B., Gharabaghi, A. BCI and robotics framework for stroke rehabilitation 4th International BCI Meeting, June 2010 (talk)

PDF Web [BibTex]

Imitation and Reinforcement Learning

Kober, J., Peters, J.

IEEE Robotics and Automation Magazine, 17(2):55-62, June 2010 (article)

Abstract

In this article, we present both novel learning algorithms and experiments using the dynamical system MPs. As such, we describe this MP representation in a way that it is straightforward to reproduce. We review an appropriate imitation learning method, i.e., locally weighted regression, and show how this method can be used both for initializing RL tasks as well as for modifying the start-up phase in a rhythmic task. We also show our current best-suited RL algorithm for this framework, i.e., PoWER. We present two complex motor tasks, i.e., ball-in-a-cup and ball paddling, learned on a real, physical Barrett WAM, using the methods presented in this article. Of particular interest is the ball-paddling application, as it requires a combination of both rhythmic and discrete dynamical systems MPs during the start-up phase to achieve a particular task.

PDF Web DOI Project Page [BibTex]

Kober, J., Peters, J. Imitation and Reinforcement Learning IEEE Robotics and Automation Magazine, 17(2):55-62, June 2010 (article)

PDF Web DOI Project Page [BibTex]

Using Model Knowledge for Learning Inverse Dynamics

Nguyen-Tuong, D., Peters, J.

In Proceedings of the 2010 IEEE International Conference on Robotics and Automation (ICRA 2010), pages: 2677-2682, IEEE, Piscataway, NJ, USA, 2010 IEEE International Conference on Robotics and Automation (ICRA), May 2010 (inproceedings)

Abstract

In recent years, learning models from data has become an increasingly interesting tool for robotics, as it allows straightforward and accurate model approximation. However, in most robot learning approaches, the model is learned from scratch disregarding all prior knowledge about the system. For many complex robot systems, available prior knowledge from advanced physics-based modeling techniques can entail valuable information for model learning that may result in faster learning speed, higher accuracy and better generalization. In this paper, we investigate how parametric physical models (e.g., obtained from rigid body dynamics) can be used to improve the learning performance, and, especially, how semiparametric regression methods can be applied in this context. We present two possible semiparametric regression approaches, where the knowledge of the physical model can either become part of the mean function or of the kernel in a nonparametric Gaussian process regression. We compare the learning performance o f these methods first on sampled data and, subsequently, apply the obtained inverse dynamics models in tracking control on a real Barrett WAM. The results show that the semiparametric models learned with rigid body dynamics as prior outperform the standard rigid body dynamics models on real data while generalizing better for unknown parts of the state space.

Jan Peters

2010

2010

2009

2009

2008

2008

Latest News

Links

Contact Us