Kun Chu

Research Interests

My research aims to develop and validate core algorithmic frameworks for
Embodied Intelligence, centering on reinforcement learning, multi-modal
foundation models, and large-scale general-purpose models tailored for
robotics (e.g., World Models, Vision-Language-Action Models). Within the
field of humanoid robotics, my work concentrates on two fundamental aspects:

1) Brain: Leveraging LLMs for high-level reasoning and long-horizon
planning in bimanual manipulation.

2) Cerebellum: Acquiring robust low-level bimanual manipulation skills
through robotic foundation models.

By integrating such a top-down framework, I aim to bridge cognition and
control, advancing the autonomy and versatility of intelligent humanoid
systems.

For more information, please visit my personal website at:

https://kchu.github.io/

Publications