Julen Urain

I am fourth year PhD student in Robot Learning at TU Darmstadt (Germany) in the Intelligent Autonomous Systems lab, advised by Prof. Jan Peters. In 2022-2023, I was a research intern in Nvidia's Robot Learning Lab lead by Prof. Dieter Fox. My research interests lie in the interplay of robotics and machine learning. In particular I explore the combination of fields such as generative modelling, motion planning, optimal control, deep learning, Riemannian geometry and physics. My primary goal is to find out the algorithmic foundations for intelligent robot behaviors. My research vision entails studying the optimal integration of the structured requirements for robotics systems with learning and optimization. Much of my previous work has focus on the adaptation of the advances in generative modelling to represent robot motions. This lead me to design motion generation architectures and explore the connections between Imitation Learning to Generative Modelling.

I am looking for robotics research positions for 2024 onwards. Also open for potential collaborations or chatting about common interests. So do not hesitate in contacting me :)


  • (02-06-2023) We won Best Paper Award in Geometric Representations Workshop at ICRA 2023 for our work on SE(3) DiffusionFields
  • (28-04-2023) I am a R:SS Pioneer! A 30 member strong-cohort of top early robotics researchers (%22 acceptance)
  • (06-03-2023) Accepted our IJRR paper on Composable Energy Policies
  • (17-01-2023) Two papers accepted for ICRA 2023: SE(3) DiffusionFields and Hierarchical Policy Blending
  • (31-10-2022) Started as Research Intern in Nvidia's Robot Learning Lab.
  • (13-09-2022) Accepted in RA-L our work on Learning Stable Vector Fields on Lie Groups

Research Highlights

Benchmarking Structured Policies and Policy Optimization for Real-World Dexterous Object Manipulation

Robotics and Automation Letters (RA-L) 2021