Jannik Zürn

Sr. Machine Learning Engineer @ Parallel Domain.

me.jpg

My name is Jannik Zürn. I am a Senior Machine Learning Engineer at Parallel Domain.

👨‍🎓 Short Bio

I am an AI researcher and engineer passionate about the intersection of robotics and simulation. My work focuses on building the sophisticated simulation environments and generative models necessary for the next generation of autonomous systems.

Current & Recent Experience

  • Parallel Domain: Currently focusing on 3D Gaussian Splatting (3DGS) and diffusion models to create high-fidelity synthetic environments for AI training.

  • Wayve: Previously worked within the Neural Simulation team, developing AI-driven world models to accelerate autonomous driving development.

🎓 Background

I hold a degree in Mechanical Engineering from the Karlsruhe Institute of Technology (KIT), specializing in Computational Mechanics and Robotics. My early career included software and research internships at ANSYS (CFD for internal combustion) and the social robotics startup Mayfield Robotics. Later, I completed a PhD in robotics and AI at the University of Freiburg, where I focused on self-supervised learning for urban navigation.

🔬 Research Interests

My PhD research focuses on bridging the gap between Robotics and Deep Learning to enable robust autonomous urban navigation. I am particularly interested in multi-modal, self-supervised learning (SSL), which allows robots to interpret complex environments without the need for expensive, manual data labeling.

Core Focus Areas:

  • Robust Perception: Developing systems that remain accurate under adversarial conditions, including sensor noise, uncertainty, and occlusions.

  • Scalable AI: Using self-supervision to bypass the “labeling bottleneck,” making the transition from simulation to real-world deployment more efficient.

  • Sensor Fusion: Leveraging diverse sensor modalities to build a more holistic and resilient understanding of dynamic urban environments.

You can find a complete list of my works on Google Scholar.

You can find my CV here.

News

Nov 2, 2023 🎉 Our paper AutoGraph: Predicting Lane Graphs from Traffic Observations was accepted to Robotics and Automation Letters (RA-L) and will be presented at ICRA 2024 in Yokohama!
Feb 27, 2023 🎉 Our paper Learning and Aggregating Lane Graphs for Urban Automated Driving was accepted at CVPR2023 - Vancouver, Canada! 🍁
Dec 21, 2022 I had the pleasure to present our brand-new paper TrackletMapper: Ground Surface Segmentation and Mapping from Traffic Participant Trajectories at CoRL 2022 in Auckland, New Zealand! 🥝

Selected Publications

  1. Wayvescenes101: A dataset and benchmark for novel view synthesis in autonomous driving
    Zürn, Jannik, Gladkov, Paul, Dudas, Sofı́a, Cotter, Fergal, Toteva, Sofi, Shotton, Jamie, Simaiaki, Vasiliki, and Mohan, Nikhil
    arXiv preprint arXiv:2407.08280 2024
    PDF Website
  2. AutoGraph: Predicting Lane Graphs from Traffic Observations
    Zürn, Jannik, Posner, Ingmar, and Burgard, Wolfram
    arXiv preprint arXiv:2306.15410 2023
    Bib PDF Website
  3. Learning and Aggregating Lane Graphs for Urban Automated Driving
    Büchner, Martin,  Zürn, Jannik, Todoran, Ion-George, Valada, Abhinav, and Burgard, Wolfram
    arXiv preprint arXiv:2302.06175 2023
    Bib PDF Website
  4. TrackletMapper: Ground Surface Segmentation and Mapping from Traffic Participant Trajectories
    Zürn, Jannik, Weber, Sebastian, and Burgard, Wolfram
    In 6th Annual Conference on Robot Learning 2022
    Bib PDF Website