I am an assistant professor (full-time mentor) at Shanghai Innovation Institute (SII), aiming to better understand, control and design complicated physical systems by AI. Our lab specifically focuses on AI for fusion (maybe the most exciting future energy that may lead human civilization to the next level) as our primary application domain, but is also interested in other physical systems such as quantum computing and brain-computer interface. Our research aims to answer three keys questions in these physical systems:

  1. Dynamics modeling (or using a more fancy word, world model): How will the system evolve under different states, control actions and environment context? Especially under extreme physical conditions such as multi-modal sensor inputs, partial observability, noise and missing values in observations, which may be rare in daily life scenarios but common in complicated physical systems.
  2. Control: How to learn a generalist controller of the complicated system efficiently by interacting with the world model learned above?
  3. Design: How should we design the system such that it can operate in a more efficient way with higher performance upper bound?

We believe AI can provide an integrated framework to tackle these challenges more efficiently and effectively than even human experts by learning and searching in the high-dimension operation space of these complicated systems. To achieve these goals, our lab investigates a broad range of AI techniques, ranging from world models (large-scale pretraining, generative models, physical AI, etc.) for dynamics modeling, to reinforcement learning, MPC for control, and all kinds of optimization methods (AI agent, generative models, Bayesian optimization, evolution algorithms, etc.) for design.

Our lab is actively looking for self-motivated PhD students, intern students, and full-time research engineers. No matter you have AI or physics/fusion/plasma background, welcome to email me to chat about potential opportunities!

Email / Google Scholar / Github / Zhihu 知乎

Education

Work Experience

Publications