Primary Focus
VLM · VLA · VLN · Multi-Robot Collaboration
Also Working On
Simulation · Humanoid Robot Control · RL Training · Motion Retargeting · Expressive Motion
机器人与具身智能方向：主要关注视觉语言模型、视觉语言动作、视觉语言导航与多机协同；同时从事仿真、人形机器人运控、强化学习训练、动作重定向与动作表演。
Robotics Laboratory, Harbin Institute of Technology (HIT)
2026.08 – Present · expected through 2027.07
哈尔滨工业大学机器人实验室 · 2026.08—至今（预计至 2027.07）
- Vision-language robotics: Vision-Language Models (VLM), Vision-Language-Action (VLA), and Vision-Language Navigation (VLN).
- Multi-robot systems: collaboration, coordination, and task execution across multiple robots.
- Humanoid robotics: whole-body control, RL policy training, motion retargeting, and expressive action performance.
- Simulation: developing and validating robot behaviors in simulation.
| Project | Area | Highlights |
|---|---|---|
| OmniXtreme | High-dynamic humanoid motion tracking | Paper · Project page & demos · pretrained policy and sim-to-sim evaluation |
| PM-V2 RL | Humanoid locomotion with MuJoCo + PPO | EngineAI PM-V2 training setup, observation/action design, reward function, and training scripts |
| BrewBot | Robotic manipulation simulation | PyBullet UR5e · RGB-D cup detection · IK and trajectory planning · FastAPI + Streamlit |
Python · C++ · ROS · MuJoCo · PyBullet · Stable-Baselines3 · FastAPI