Sami Azirar | University of Bonn

About Me

I am a Ph.D. student at the Robotics Perception and Learning Lab at the University of Bonn. During my PhD, I aim to answer the question of how we can train robots using web videos. For this, I want to use video generation models as world simulators. I aim to enable generalization in robotics by scaling across enviroments, tasks, and embodiments!

I am also associated with LAMARR.

Feel free to contact me for collaborations

Research Interests

World Models: Video generation, physical plausibility, temporal coherence
Robotics: Trajectory learning, cross-embodiment transfer, generalization

News

[Jul. 2026] New paper on arXiv: ContactFlow, a video action conditioning that transfers across embodiments.
[Jul. 2026] Our paper SYMBOLIZER on symbolic model-free task planning with VLMs is accepted to RA-L!
[Jun. 2025] I had much fun at the ETH Robotic Summer School!
[Feb. 2025] I started as a PhD Student at the RPL Lab Uni Bonn!
[Feb. 2024] I started my Master Thesis about Task Planning at the RWTH RLeap in Aachen.
[Apr. 2024] Our Paper about using LLMs as a intelligent query system is accepted to ICASI.
[Jun. 2023] I started as a Graduate Research Fellow at Ontario Tech/ NRC Canada.

Publications

RA-L

SYMBOLIZER: Symbolic Model-free Task Planning with VLMs

Sami Azirar*, Zlatan Ajanovic*, Hermann Blum
* equal contribution

IEEE Robotics and Automation Letters (RA-L), 2026.

PDF Accepted to RA-L

arXiv

ContactFlow: A video action conditioning that transfers across embodiments

Sami Azirar, Enrico Pallotta, Jan Nogga, Jürgen Gall, Sven Behnke, Hermann Blum

arXiv preprint arXiv:2607.26579, 2026.

PDF Project Page Preprint

ICASI

IQLS: Framework for leveraging metadata to enable large language model based queries to complex, versatile data

Sami Azirar, Hossam A. Gabbar, Chaouki Regoui

International Conference on Applied System (ICASI), 2024.

PDF BibTex