Skip to content

About

I am Yaozeng Huang (黄耀增), a third-year undergraduate at Xinjiang University, majoring in Computer Science. I'm originally from Zhoushan, Zhejiang.

My research interests sit at the intersection of embodied AI and reinforcement learning:

  • VLA models for contact-rich manipulation — at Tsinghua SIGS, working on force-aware VLA, geometric failure modes, and RGB-D privileged-geometry distillation.
  • RL for dynamic-obstacle navigation — collaborating with Eastern Institute of Technology, Ningbo on extending local planning from static to dynamic settings.
  • Legged locomotion — at the XJU Innovation Lab, training PPO / N-P3O policies and pushing through MuJoCo sim-to-sim into sim-to-real on quadrupeds.

I believe evidence on real hardware is more convincing than simulation curves, which is why over the past two years I've spent serious time on robotics competitions — ROBOCON and the National Electronic Design Contest — picking up 9 national-level and 3 provincial-level awards along the way.

Contact

Email13325905201@163.com
GitHub@hyyyyyyz
Based inÜrümqi, Xinjiang, China
HometownZhoushan, Zhejiang
CommunityMember of the Xbotics embodied-AI open-source community

If your work overlaps with embodied AI, VLA models, or robot learning, I'd love to hear from you — I am open to remote collaborations and open-source contributions.

About this site

This is my personal academic homepage, built with VitePress and deployed to GitHub Pages. Source code: hyyyyyyz/hyyyyyyz.github.io.

The home page is updated continuously as research progresses; the Notes section contains my personal study notes (written in Simplified Chinese) on reinforcement learning, ROS 2, MuJoCo, Git, Docker, and VitePress.

The footprint of this site is intentionally small — one font family, one accent colour, one source of truth. As an undergrad, I'd rather spend my hours doing research than tweaking a portfolio.

Last updated:

Built with VitePress · Source on GitHub