Kun Chu

Kun Chu

All about learning.

All about me

PhD student in Computer Science, University of Hamburg, Knowledge Technology (WTM) group, advised by Prof. Stefan Wermter.

I am deeply interested in the critical methodologies driving Embodied Intelligence, particularly Reinforcement Learning, Large Language Models (LLMs), and Robotics. In humanoid robot development, three key components shape their capabilities: the Brain, the Cerebellum, and the Body. My current research focuses on two of these aspects:

  • Brain – Leveraging LLMs for planning in long-horizon bimanual robotic tasks.

  • Cerebellum – Developing bimanual manipulation skills using foundation models.

Through my work, I aim to bridge high-level reasoning and planning with low-level motion control, pushing the boundaries of intelligent robotic systems.

The Future of Embodied Intelligence is Now.

All about world models

In May 2026, I initiated the WTM World Model Seminar, an ongoing seminar series in our group where colleagues present and discuss the latest work on world models, share ideas across projects, and explore how these advances can inform our own research. Anyone interested in world models is warmly welcome to join us — feel free to reach out or simply drop by!

All about news

🔥 [2026.03] I was honored to take part as an invited participant in NII Shonan Meeting Seminar No.235, LLM-guided Synthesis, Verification, and Testing of Learning-Enabled CPS, held at the Shonan Village Center in Japan.

🔥 [2026.01] Our paper, RankCut: A Ranking-Based LLM Approach to Extractive Summarization for Transcript-Based Video Editing, has been accepted to IUI 2026. This project was co-supervised with the Adobe Research team. Congrats, Sana!

[2025.05] I was honored to attend IJCNN 2025 in Rome, thanks to the generous support of the IEEE CIS Travel Grant.

[2025.04] Our paper, LLM-iTeach: LLM-based Interactive Imitation Learning for Robotic Manipulation, has been accepted for an oral presentation at IJCNN 2025! This work was done by Jonas under my supervision. Congrats, Jonas!

[2025.03] Our paper, LLM+MAP: Bimanual Robot Task Planning using Large Language Models and Planning Domain Definition Language, is now available online on arXiv. Please check the video as well on Youtube.

Earlier news (2023–2024) — click to expand

[2024.10] We have open-sourced the code for the LABOR Agent on GitHub.

[2024.09] Our paper Large Language Models for Orchestrating Bimanual Robots has been accepted by Humanoids 2024, and was presented in Nancy, France that November.

[2024.07] Our application to Researcher Access Program of OpenAI has been accepted, and we were awarded 5,000 USD in API credits!

[2024.02] I was glad to serve as a technical committee member of the Workshop on Human-aligned Reinforcement Learning for Autonomous Agents and Robots at the 2024 IEEE International Conference on Robotics and Automation (ICRA 2024).

[2024.02] Our paper Enhancing Zero-Shot Chain-of-Thought Reasoning in Large Language Models through Logic has been accepted at LREC-COLING 2024.

[2023.11] Our paper Accelerating Reinforcement Learning of Robotic Manipulations via Feedback from Large Language Models has been accepted as an oral presentation at the 7th Conference on Robot Learning (CoRL 2023) Workshop on Bridging the Gap between Cognitive Science and Robot Learning in the Real World: Progresses and New Directions.

All about academic service

Journal Reviewer

Conference Reviewer

  • IROS, ICRA, Humanoids
  • IUI, IJCNN
  • EAI MobiQuitous

Technical Program Committee Member

All about my students

It is a privilege to work with these talented students. Their projects span the topics I care about most — agentic systems, world models, and robot learning.

  • Sana — RankCut: Extractive Summarization for Transcript-Based Video Editing (IUI 2026, co-supervised with Adobe Research)
  • Jonas — LLM-iTeach: Interactive Imitation Learning for Robotic Manipulation (IJCNN 2025, oral)
  • Yue & Jingfan — Agentic LLM systems for education: structured learner modeling and grounded generation (in preparation)
  • Aamir — LLM+PDDL for abstractive world models in dynamic environments (in preparation)
  • Asihati — World models for whole-body control of humanoid robots (in preparation)

Glad to discuss anything! Please contact me at kun.chu at uni-hamburg dot de