Hi, I'm
Haotian Xue
I explore
I am a graduate student in Software Engineering at Sichuan University. My interests include multi-agent reinforcement learning, world models, autonomous systems, robotics, intelligent transportation, and open-source software. I use this website to share my projects, research notes, technical tutorials, and reflections on learning and development.
Featured Projects
Selected research, engineering, and open-source work
Llama WebGPT
A web frontend built with Flask, Vue, JavaScript, and CSS, integrated with a fine-tuned LLaMA large language model.
Software Project
A software engineering course design project completed during my undergraduate study at the Civil Aviation University of China.
Tech Stack
Tools I use for research, development, and experimentation
Languages
Python
C/C++
Java
Markdown
AI & Robotics
PyTorch
ROS
Jupyter
Development
Git
GitHub
Linux
Web
Flask
Vue.js
JavaScript
Education
Graduate Student in Software Engineering
Sichuan University
Graduate study in Software Engineering, with interests in multi-agent
intelligence, autonomous systems, robotics, and world models.
B.Eng. in Computer Science and Technology
Civil Aviation University of China
Undergraduate study in Computer Science and Technology, including
software engineering, systems programming, networking, and open-source projects.
Selected Publications
Latest Posts
$\pi_0$ 精读：从 Action Token 到 Flow Matching，VLA 开始学习灵巧控制
快速读懂 $\pi_0$：它如何在 RT-2 与 OpenVLA 建立的 VLA 范式上，引入 PaliGemma、Action Expert、Flow Matching、Action Chunking 与跨机器人预训练，把视觉语言知识进一步连接到连续、高频、灵巧的机器人控制。
DreamerV3 精读：在世界模型中想象未来，再学习策略
快速读懂 DreamerV3：RSSM 如何学习潜空间世界模型，Actor-Critic 如何在 imagined trajectories 中训练，以及 symlog、two-hot、KL balancing、free bits 与 return normalization 为什么让一套超参数可以跨 150+ 任务工作。
Get In Touch
Research collaboration, open-source development, and technical discussions
Connect
I am interested in discussing artificial intelligence, multi-agent systems, autonomous systems, robotics, world models, and open-source projects. Feel free to contact me by email or through GitHub.
✓ Copied to clipboard!