printlogo
http://www.ethz.ch/index_EN
Institute for Dynamic Systems and Control
 
print
  

Hübel, Nico

Nico Huebel

ETH Zürich
Nico Huebel
Inst. Dynam. Syst. u. Regelungst.
ML K 35
Sonneggstrasse 3
8092 Zuerich

Phone: +41 44 632 73 68
E-Mail: 

Bio

Nico Huebel is pursuing his PhD in engineering with Professor Raffaello D'Andrea at the Institute for Dynamic Systems and Control (IDSC) at ETH Zurich (Switzerland). His research interests are in the area of adaptive and learning controllers for autonomous systems and how learned knowledge can be shared between robots. Nico is currently working on RoboEarth, an EU project that has the goal to create an online database for robots to share knowledge with each other. He is responsible for maintaining the software and hardware setup of the KUKA Lightweight Robot testbed. Beside his research interests he enjoys teaching and passing on his passion about mathematics and robotics to people of all ages.

Nico received his Master's degree (Diplom-Ingenieur) in Engineering Cybernetics from the University of Stuttgart (Germany) in 2010 with a focus on cognitive robotics. In 2007 he spent 7 months at the Tokyo Institute of Technology working on Coverage Control with Information Decay for Mobile Sensor Networks with Professor Masayuki Fujita. He was also working for KUKA Robotics, where he developed novel strategies for force torque control and adaptive control for industrial robots. As a member of the Middle Size League RoboCup team he was working on hardware development and motion control. The team consistently ranked among the best three teams in various national and international competitions including the first runner-up in the 2006 World Championships. Nico worked as a teaching and research assistant throughout his studies and was awarded a scholarship from the German Academic National Foundation.

Research Interests

Nico's research can be broken down into the following areas:

Robotic Calligraphy

Calligraphy is the art of beautifully writing with a brush. Beside its aesthetic and artistic challenges, calligraphy requires graceful and complex motions. That makes it a challenging problem for machine learning. Humans solve the problem by practicing the same strokes over and over again. We have developed a learning algorithm that also iteratively improves its performance over several trials. The performance can be seen in the following video:

Although thousands of Chinese and Japanese characters exist, they can be written using a limited set of single strokes. This property is useful when building up a database of characters using the strokes as motion primitives. However, this property can also be exploited for automatically creating the task description for writing a character. An image processing algorithm was created that uses the database of learned motion primitives to automatically create a task description for drawing a newly presented character. This algorithm not only produces a sequence of previously learned motion primitives, but also adapts them in order to achieve a more accurate result. More information can be found on the corresponding project page.

Sharing of Kowledge between Robots

The goal of the RoboEarth project is to build a database that can be used to share knowledge between robots. I am focusing on how to efficiently represent, store and reuse trajectory based task knowledge learned by robots.

The following video shows a demonstration in which a robot with less accurate sensing capabilities benefits from the knowledge provided by a robot with more accurate sensing capabilities:

However, it is difficult for a robot to decide whether its learned knowledge is beneficial to share or if there is already similar knowledge in the database. Therefore, we developed an algorithm to find similar trajectory based knowledge and fuse such similar models in order to remove duplicate models. Please find more details on the project page.

Ontology-Based Software Configuration and Composition

For dynamic tasks, trajectory based motion primitives are not suitable. For such tasks software components, e.g. a controller, need to be configured based on the properties of the robot, the task, and the object. In addition, many robotic applications require the interaction of several software components. We have developed a way of using an ontology to configure and compose several software components to interact with each other in order to fulfill a task like the following:

Invited Talks

Tokyo Institute of Technology, Tokyo, Japan, Robotic Calligraphy – A Learning Approach, Systems and Control Seminar, Nov 2013

KUKA Laboratories GmbH, Augsburg, Germany, Selected Demonstrations of RoboEarth and the KUKA LWR, Sep 2013

Publications (Google Scholar)

S. Mueller, N. Huebel, M. Waibel, R. D'Andrea, Robotic calligraphy - Learning how to write single strokes of Chinese and
Japanese characters
, in IEEE/RSJ International Conference of Intelligent Robots and Systems, 2013, accepted.

G. Reinhart, T. Martens, S. Krotil, N. Hübel, A new approach for an efficient and flexible control of biogas production in a full scale plant, Poster (peer-reviewed) at the World Congress on Anaerobic Digestion 2013 (pdf) | (bib)

F. Ramos, M. Gajamohan, N. Huebel, R. D’Andrea, Time-Optimal Online Trajectory Generator for Robotic Manipulators, Technical Report, 2013 (pdf) | (bib)

N. Huebel, E. Mueggler, M. Waibel, R. D'Andrea, Towards robotic calligraphy, in IEEE/RSJ International Conference of Intelligent Robots and Systems, 2012, pp. 5165 - 5166. (pdf) | (bib)

N. Hübel, S. Hirche, A. Gusrialdi,T. Hatanaka, M. Fujita, O. Sawodny, Coverage Control with Information Decay in Dynamic Environments, in 17th IFAC World Congress, 2008, pp. 4180 - 4185. (pdf) | (bib)

N. Hübel, S. Hirche, A. Gusrialdi, T. Hatanaka, M. Fujita, O. Sawodny, "Infinite Time Coverage Control with Information Decay for Mobile Sensor Networks," in 36th SICE Annual Control Theory Symposium, 2007, pp. p. 23-26. (pdf) | (bib)

Teaching

Dynamic Programming and Optimal Control, Fall 2012, Teaching Assistant
Dynamic Programming and Optimal Control, Fall 2011, Teaching Assistant
Dynamic Programming and Optimal Control, Fall 2010, Teaching Assistant

Supervision

Andreas Forster: Robotic calligraphy – Automatic stroke segmentation (Bachelor Thesis)
Andreas Forster: Survey of stroke segmentation algorithms (Studies on Mechatronics)
Thomas Konrad: Creating an Ontology for Auto-Configuring Controllers (Semester Project)
Nathanael Wettstein: Balancing a Ball on a Plate using Stereo Vision (Master Thesis)
Nathanael Wettstein:Modelling and Simulation of a ball on a plate (Semester Project)
Christoph Zumbrunn: Robotic Calligraphy: Automatic Identification of Hyper-Parameters for the Learning Algorithm (Bachelor Thesis)
Christoph Zumbrunn: Survey of Spline Fitting Methods (Studies on Mechatronics)
Andreas Bircher: Fusion Component for Trajectory Information (Semester Project)
Mauro Salazar: MRAC Algorithms for Mobile Robots (Bachelor Thesis)
Mauro Salazar: Survey of Model Reference Adaptive Control Algorithms (Studies on Mechatronics)
Tom Winandy: From White Box to Black Box Models (Semester Project)
Samuel Müller: Robotic Calligraphy (Master Thesis)
Daniel Eberli: Modeling and Control of a Ball Balancing System Using Stereo Vision (Master Thesis)
Kilian Schindler: Competing with Human Dexterity - Minimizing the time to go through a maze using haptic feedback (Bachelor Thesis)
Kilian Schindler: Algorithms for fast exploration (Studies on Mechatronics)
Elias Müggler: Robotic calligraphy - A robot that learns how to write Chinese calligraphy (Semester Project)
Roman Bosshard:Paddle Trajectory Generation (Semester Project)
Andreas Ritter: Model analysis and trajectory generation for a four-wheeled omni-directional ground vehicle (Bachelor Thesis)

 

Wichtiger Hinweis:
Diese Website wird in älteren Versionen von Netscape ohne graphische Elemente dargestellt. Die Funktionalität der Website ist aber trotzdem gewährleistet. Wenn Sie diese Website regelmässig benutzen, empfehlen wir Ihnen, auf Ihrem Computer einen aktuellen Browser zu installieren. Weitere Informationen finden Sie auf
folgender Seite.

Important Note:
The content in this site is accessible to any browser or Internet device, however, some graphics will display correctly only in the newer versions of Netscape. To get the most out of our site we suggest you upgrade to a newer browser.
More information

© 2014 ETH Zurich | Imprint | Disclaimer | 26 November 2013
top