Intelligent Systems

Group News

Branco Weiss Lecture

  • 14 December 2023

Public lecture on "(un)"predictability - a journey through elements of chaos and control theory, quantum mechanics, machine learning, and how humans deal with (un)predictability.


Presentation at NeurIPS

  • 24 November 2023

Our work "Online Learning under Adversarial Nonlinear Constraints" was accepted to NeurIPS23.


Roboter entdecken die Welt

  • 06 October 2023

Forschende vermitteln Robotern eine Art Körpergefühl

Roboter können den Menschen heute bereits bei manchen alltäglichen Aufgaben unterstützen. Doch unbekannte Umgebungen oder auch kleine Abweichungen in den Aufgaben, auf die sie trainiert sind, überfordern sie. Damit sie rascher lernen, sich auf Neues einzustellen, entwickeln die Forschungsgruppen von Michael Mühlebach und Jörg Stückler am Max-Planck-Institut für Intelligente Systeme in Tübingen neue Trainingsmethoden für die Maschinen. Ihre Roboter müssen sich dabei auch im Pingpong oder Bodyflying bewähren.

Michael Muehlebach Jörg Stückler Dieter Büchler Hao Ma


L4DC conference

  • 13 June 2023

Two articles from our group will be presented at the Learning for Dynamics and Control Conference.

Congratulations to Guanchun, Jan, Philip, and Hao for having their articles accepted at the Learning for Dynamics and Control Conference.


UAI Conference

  • 13 June 2023

Our article will be presented at the Conference for Uncertainty Quantification in Artificial Intelligence.

Congratulations to Klaus and Julius!


COLT conference

  • 12 June 2023

Our article on stochastic gradient algorithms on manifolds has been accepted to the Conference on Learning Theory 2023.

Congratulations to Sholom and Daniil!


Presentation at NeurIPS

  • 28 November 2022

Our work "Sampling without replacement leads to faster rates in finite-sum minimax optimization" was accepted at NeurIPS 2023.


The One-Wheel Cubli

  • 15 August 2022

A new balancing robot, see https://www.youtube.com/watch?v=LM74EBbetGU .


Presentation at RSS

  • 20 June 2022

We will present our work on learning-based iterative control at RSS 2022.


Invited talk at the NCCR Symposium

  • 09 May 2022

The NCCR Symposium is biannual symposium covering a broad range of topics in relation to NCCR Automation research.