• Nutan Chen, Alexej Klushyn, Richard Kurle, Xueyan Jiang, Justin Bayer, Patrick van der Smagt (2017)
    Metrics for Deep Generative Models
    arXiv [www]
  • Maximilian Karl, Maximilian Soelch, Philip Becker-Ehmck, Djalel Benbouzid, Patrick van der Smagt, Justin Bayer (2017)
    Unsupervised Real-Time Control through Variational Empowerment
    arXiv [www]
  • Maximilian Karl, Maximilian Soelch, Justin Bayer, Patrick van der Smagt (2017)
    Deep Variational Bayes Filters: Unsupervised Learning of State Space Models from Raw Data
    International Conference on Learning Representations (ICLR) [pdf|bibtex]
  • Baris Kayalibay, Grady Jensen, Patrick van der Smagt (2017)
    CNN-based Segmentation of Medical Imaging Data
    arXiv [www|bibtex]
  • Nutan Chen, Maximilian Karl, Patrick van der Smagt (2016)
    Dynamic Movement Primitives in Latent Space of Time-Dependent Variational Autoencoders
    Proc. 16th IEEE-RAS International Conference on Humanoid Robots [pdf|bibtex]
  • See for previous related papers.

about us

AI Research in the Volkswagen Group Munich Data:Lab conducts fundamental machine-learning research, working towards artificial intelligence. The open-access research group exploits and develops methodologies around deep learning, variational inference, and time series modelling. We validate our methodologies in sensorised robotics and control settings.

Disseminating our scientific results is key to our mission. We will keep you updated through our publications, an upcoming blog, and social media posts.

Deep Learning
and Robotics Challenge

The Volkswagen Group AI Research Lab—a part of the Data:Lab in Munich—invites students to join the 2017 Deep Learning and Robotics Challenge.

In this challenge, which is scheduled to take place from September 11 to October 13, 2017, you will collaborate with your peers in small groups to solve a robotics challenge. Your methodologies should be based on deep neural networks, which you are to develop within the TensorFlow framework. Python is our preferred language.

All groups will compete to have their robot solve the set task as efficiently as possible. The challenge will be crowned with a ceremony at the GTC Europe (Oct. 10--12, Munich) for the winning team. All code that you will develop within this challenge will be open-sourced. Note that your major input will be related to machine learning, and less so to robotics.

The DLRC is set up in a collaboration between Volkswagen Group and NVIDIA. In this collaboration, we will provide you with a compute infrastructure based on DGX-1, 1080s and Jetsons. You need to bring your own laptop to access our compute clusters.

Registration was closed on August 1, 2017.

Follow the students' progress on their blog


This website is intended as an informational platform, describing research output by the AI Research Group within the Data:Lab in Munich. The Data:Lab is part of Volkswagen Group.

None of the information published on this website officially represents Volkswagen Group or the Data:Lab, and the opinions herein do not necessarily represent the opinions of either party.

Responsible for the contents of this website is

Patrick van der Smagt
Director of AI Research
Volkswagen Group, Data:Lab
Ungererstr. 69
80805 München


The content of this website was created with great care, but correctness is not guaranteed. Pursuant to §7 par. 1 of TMG (German Telemedia Act), we are responsible under the general laws for our own content on this website. However, according to §§ 8 to 10 of TMG, we are not obligated to monitor any third-party information transmitted or stored on our website or to investigate circumstances indicating an illegal activity. This does not affect the obligations under the general laws to remove or block the use of information. However, liability in this respect is only possible from the time we become aware of an actual infringement of law. Once we become aware of such legal violations, we shall remove the respective contents immediately.

Liability for links

Our service includes links to external websites but we have no influence over their content. Therefore, we cannot accept responsibility for these third-party contents. In each case, it is the respective provider or operator of the linked websites being responsible for their contents. Continuous monitoring of the contents of the linked pages cannot be reasonably expected unless there are concrete reasons for suspecting legal violations. Once we are aware of illegal content, we shall remove the respective links immediately.


The contents and works created by the page operators on these pages are subject to German copyright law. The duplication, editing, dissemination and all kinds of use beyond the constraints of copyright law require written approval by the respective author or creator. Downloads and copies of these pages are only for private and not for commercial use. Insofar as the contents on this website were not created by the operator, third-party copyrights are respected. Should you nevertheless notice a copyright infringement, please inform us accordingly. Once we become aware of violations of rights, we shall remove the respective contents immediately.

Data protection

It is generally possible to use our website without providing personal data. Insofar as we collect personal data via this website (for example, name, address or e-mail addresses), this is, as far as possible, always on a voluntary basis. Your personal data will not be shared with third parties without your express permission. We wish to remind you that there are inherent security risks in transmitting data (for example, email communication) via the Internet. It is not possible to ensure that data is completely protected against unauthorised access by third parties. Third parties are expressly prohibited from using contact details published herein, as required under the “Impressum” (legal notice) requirement of German law, for the purpose of sending advertising or informational material which has not been specifically requested. Where spam e-mails or other unsolicited advertising information is sent, the operators of the relevant websites specifically reserve the right to take legal action.