I am now a full professor and chair at MU Leoben, see my new homepage ...

Elmar Rueckert

Elmar Rueckert joined the IAS team in March 2014 as postdoctoral researcher and research group leader. He investigates computational models of the motor control system and validates them in robotic applications. A strong emphasis is put on neurorobotics (N), deep networks and machine learning (D), neuro-prostheses applications with brain-machine interfaces (P), and on computational neural models of human motor control and learning (H).

Before coming to Darmstadt, Elmar did his Ph.D. at the Graz University of Technology (TUG) under the supervision of Wolfgang Maass. Elmar started his Ph.D. studies in February 2010 and passed its defense with distinction in February 2014. During his Ph.D., he worked on the AMARSi project, where he developed novel reinforcement learning algorithms for motor planning using probabilistic inference, biologically inspired movement primitive representations based on muscle synergies, and investigated how networks of spiking neurons can solve motor control and motor planning problems. During his Ph.D., he also collaborated with Marc Toussaint, Andrea d'Avella, and Thomas Schack. His Thesis, "On Biologically inspired motor skill learning in robotics through probabilistic inference" concentrated on probabilistic inference for motor skill learning and on learning biologically inspired movement representations. Elmar defended his PhD thesis in February 2014.

Elmar was born in Unterpremstätten, Austria. He received his qualification for university entrance at the technical high school for electronic engineering and informatics, HTL Graz Gösting. Before doing his PhD, he finished his studies in telematics at the TUG in the year 2010. From 2012 to 2014, Elmar gave the data structures and algorithms lecture, where he received an outstanding good reputation from his students. During his Ph.D., he also supervised several student's projects on machine learning and robotics.

Research interests

Neurorobotics

• Probabilistic inference for motor planning an control
• Movement representations for motor skill learning
• Reinforcement Learning for robotics

Deep Networks and Machine Learning

• Spiking Neural Networks for Motor Control, Planning and Learning
• Meta- or structure learning for robotics

Computational Neural Models of Human Motor Control

• Modeling Human Motor Learning
• Biologically inspired motor skill learning approaches

Neuroprostheses and Brain-Machine Interfaces

• Novel training schemes for Brain-Computer Interfaces
• Cybathlon related signal processing and pattern recognition
  
  

Key References

1. Rueckert, E.; Camernik, J.; Peters, J.; Babic, J. (2016). Probabilistic Movement Models Show that Postural Control Precedes and Predicts Volitional Motor Control, Nature PG: Scientific Reports, 6, 28455.   Download Article [PDF]   BibTeX Reference [BibTex]

  Matlab Code Probabilistic Trajectory Model

2. Rueckert, E.; Kappel, D.; Tanneberg, D.; Pecevski, D; Peters, J. (2016). Recurrent Spiking Networks Solve Planning Tasks, Nature PG: Scientific Reports, 6, 21142, Nature Publishing Group.   Download Article [PDF]   BibTeX Reference [BibTex]

  Matlab Code Neural Network Framework 

3. Rueckert, E.; Mundo, J.; Paraschos, A.; Peters, J.; Neumann, G. (2015). Extracting Low-Dimensional Control Variables for Movement Primitives, Proceedings of the International Conference on Robotics and Automation (ICRA).   Download Article [PDF]   BibTeX Reference [BibTex]

4. Rueckert, E.A.; Neumann, G.; Toussaint, M.; Maass, W. (2013). Learned graphical models for probabilistic planning provide a new class of movement primitives, Frontiers in Computational Neuroscience, 6, 97.   Download Article [PDF]   BibTeX Reference [BibTex]

5. Rueckert, E.A.; d'Avella, A. (2013). Learned parametrized dynamic movement primitives with shared synergies for controlling robotic and musculoskeletal systems, Frontiers in Computational Neuroscience, 7, 138.   Download Article [PDF]   BibTeX Reference [BibTex]

6. Rueckert, E.A.; Neumann, G. (2012). Stochastic Optimal Control Methods for Investigating the Power of Morphological Computation, Artificial Life.   Download Article [PDF]   BibTeX Reference [BibTex]

For all publications please see his Publication Page

Videos

(:youtube XAj944lDPPQ:)

(:youtube 2rxR2TwPVxw:)

Software

• Supplementary Matlab Code to Recurrent Spiking Networks Solve Planning Tasks, The framework contains several demo programs of different sampling methods, discrete and continuous problems and illustrates the effect of online and offline learning rules. Further, we demonstrate how reinforcement learning and imitation learning can be implemented and applied to robotic tasks.
• Matlab Code MEX-Function implementation of Locally Weighted Regression (LWR) for real-time predictions of learned models.
• Matlab Code Extracting Low-Dimensional Control Variables for Movement Primitives
• Matlab Code Robust Policy Updates for Stochastic Optimal Control
• Matlab Code Sensor Glove Matlab Mex Interface
• Matlab Code Probabilistic Model of Trajectories (a sub part of ProMPs)
• OpenSim/Simtk Matlab Interface and Biped Walker Matlab Simulator used in Rueckert, E.A.; d'Avella, A. (2013). Learned parametrized dynamic movement primitives with shared synergies for controlling robotic and musculoskeletal systems, Frontiers in Computational Neuroscience, 7, 138.
  
  

Workshops and Summer Schools

• [2016, Organized by Elmar Rueckert and Martin Riedmiller] NIPS 1.5 days workshop. Title: Neurorobotics: A chance for new ideas, algorithms and approaches.
• [2014, Invited Talk] TEDUSAR Summer School. Title: An introduction to robot learning and probabilistic movement planning.
• [2011, Organized by Gerhard Neumann and Elmar Rueckert] Two days workshop. Title: Hands-on Probabilistic Inference for Motor Control.
  
  

Current Ph.D. Students

Bachelor and Master Students

Supervised Theses and Projects at IAS

Supervised Theses and Projects at Technische Universitaet Graz