Han Zhao 赵晗

About Me

Hi! I’m Han Zhao (赵晗).I got my bachelor’s and master’s degrees in Control Science and Engineering from Beijing University of Posts and Telecommunications (BUPT) in 2020 and 2023, respectively. I am currently a first-year joint Ph.D. student in Computer Science and Technology at Zhejiang University and Westlake University (Machine Intelligence Lab, MiLAB), advised by Prof. Donglin Wang.

Research Interests

My current research interests include Embodied Artificial Intelligence, Foundation Models, Reinforcement Learning, and Robotics.

Specifically, I am interested in:

• Foundation Models for Robotics: developing efficient and effective foundation models for robotics, including multi-modal large language models and vision-language-action models to enhance the perception and decision-making capabilities of robots.

• Scalable Reinforcement Learning Algorithms: developing reinforcement learning algorithms that can effectively manage large-scale data and model capacity for robotic control. This includes methods such as offline reinforcement learning, imitation learning, and more, with the goal of enabling robots to acquire scalable and generalizable skills.

• Motion Planning and Control for Legged Robots: developing motion planning and control algorithms for legged robots, including bipedal and quadruped robots, to enable them to perform complex tasks in real-world environments.

News

• [March 22, 2024] A new paper about Cobra, an efficient multi-modal large language model, was released. Project page has been available. The paper has been featured by Hugging Face Daily Papers! Demo has been available!

• [March 20, 2024] A new paper about GeRM, a generalist robotic model with the mixture-of-experts architecture and RL training method for quadruped robot, was released. Project page has been available. Video has been available!

Publications

Preprint

Han Zhao, Min Zhang, Wei Zhao, Pengxiang Ding, Siteng Huang, Donglin Wang*, "Cobra: Extending Mamba to Multi-Modal Large Language Model for Efficient Inference". arXiv preprint arXiv:2403.14520. [paper] [project page] [zhihu] [github] [demo] [twitter@AK]

Wenxuan Song, Han Zhao, Pengxiang Ding, Can Cui, Shangke Lyu, Yaning Fan, Donglin Wang*, "GeRM: A Generalist Robotic Model with Mixture-of-experts for Quadruped Robot". arXiv preprint arXiv:2107.03575. [paper] [project page] [video]

Pengxiang Ding, Han Zhao, Zhitao Wang, Zhenyu Wei, Shangke Lyu, Donglin Wang*, "QUAR-VLA: Vision-Language-Action Model for Quadruped Robots". arXiv preprint arXiv:2107.03575. [paper]

Hongyin Zhang, Diyuan Shi, Zifeng Zhuang, Han Zhao, Zhenyu Wei, Feng Zhao, Sibo Gai, Shangke Lyu, Donglin Wang*, "RSG: Fast Learning Adaptive Skills for Quadruped Robots by Skill Graph". arXiv preprint arXiv:2107.03575. [paper]

2023

Shangke Lyu, Han Zhao, Donglin Wang*, "A Composite Control Strategy for Quadruped Robot by Integrating Reinforcement Learning and Model-Based Control". IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). [paper]

Lei Guo*, Han Zhao, "Online adaptive optimal control algorithm based on synchronous integral reinforcement learning with explorations". Neurocomputing. [paper]

Lei Guo*, Han Zhao, "Model‐free adaptive optimal control of continuous‐time nonlinear non‐zero‐sum games based on reinforcement learning". IET Control Theory and Applications. [paper]

2022

Han Zhao*, Lei Guo, Yuan Song, "System Modelling and Controller Design of a Variable Structure Two-Wheeled Robot Using Robust Adaptive Dynamic Programming". Chinese Control and Decision Conference (CCDC). [paper]

Han Zhao*, Lei Guo, "Model-free Nearly Optimal Control of Constrained-Input Nonlinear Systems Based on Synchronous Reinforcement Learning". Chinese Control Conference (CCC). [paper]

Service

Reviewer for:

Journal

• Knowledge-Based Systems (KBS)
• IET Control Theory & Applications (IET-CTA)

Conference

• IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)
• IEEE Conference on Decision and Control (CDC)