Reinforcement-Learning-Driven Gait · Full-Stack Open Source · Under ¥2,000 BOM
🚧 Demo video coming soon — stay tuned!
Getting Started · BOM · Assembly Guide · Architecture
Q1 nano is a desktop-sized servo humanoid robot that walks with a natural, fluid gait — powered by MuJoCo simulation and reinforcement learning.
Traditional servo humanoid robots are known for stiff, jerky movements. Q1 nano breaks that stereotype by training walking policies in simulation and deploying them to real hardware, achieving smooth, human-like locomotion on standard hobby servos.
| Feature | Description |
|---|---|
| 🧠 RL-Driven Gait | Smooth walking trained via reinforcement learning in MuJoCo — not hand-tuned keyframes |
| 🔧 Fully Reproducible | Complete BOM, STEP files, URDF, assembly guide — build one at home |
| 💰 Affordable | Total BOM cost under ¥2,000 (~$275 USD) |
| 📐 SolidWorks → URDF → MuJoCo → Real | End-to-end pipeline from CAD to walking robot |
| 🌍 Open Source | Hardware, simulation, training, and firmware — everything is open |
Q1-nano/
├── docs/ # Documentation
│ ├── getting_started.md # Quick start guide
│ ├── bom.md # Bill of Materials + purchase links
│ ├── assembly_guide.md # Step-by-step assembly instructions
│ └── architecture.md # System architecture overview
├── hardware/
│ ├── solidworks/ # Original SolidWorks CAD files
│ ├── step/ # STEP format (universal CAD exchange)
│ └── urdf/ # URDF model for simulation
├── simulation/
│ ├── mujoco/ # MuJoCo simulation environment
│ └── configs/ # Training & simulation configs
├── training/
│ ├── rl/ # Reinforcement learning training code
│ ├── checkpoints/ # Pre-trained model weights
│ └── scripts/ # Training & evaluation scripts
├── firmware/
│ ├── servo_control/ # Servo communication & control
│ └── main_controller/ # Main controller program
└── media/
├── demo_videos/ # Demo videos & GIFs
└── renders/ # 3D renders & photos
Note: Full quick-start instructions are under development. Check back soon!
# Clone the repository
git clone https://github.com/RobotBase/Q1-nano.git
cd Q1-nano
# Check out the docs
cat docs/getting_started.md- Repository scaffolding & README
- Upload development history & existing codebase
- URDF model & MuJoCo simulation environment
- Demo video of RL-trained walking gait
- BOM & assembly guide
- v0.1 Release — community can clone → simulate → build
- Submit to awesome-robotics lists
- Community contributions & iteration
We welcome contributions of all kinds! Whether it's fixing a typo, improving documentation, tuning RL parameters, or designing new gaits — every contribution matters.
Please read our Contributing Guide to get started.
Check out our Good First Issues for beginner-friendly tasks.
This project is licensed under the Apache License 2.0.
Q1 nano 是一款桌面级舵机人形机器人，通过 MuJoCo 仿真 + 强化学习训练步态，实现丝滑自然的行走效果。
传统舵机人形机器人步态僵硬笨拙，Q1 nano 打破了这一刻板印象——在仿真环境中训练行走策略，再部署到真实硬件，用普通舵机实现接近人类的步态。
- 🧠 强化学习驱动 — 非手调关键帧，而是 RL 训练出的自然步态
- 🔧 完全可复现 — 从 BOM 清单到组装指南，在家就能造
- 💰 低成本 — BOM 总成本 ¥2,000 以内
- 📐 全链路开源 — SolidWorks → URDF → MuJoCo → 强化学习 → 真机部署
- 🌍 全栈开放 — 硬件、仿真、训练、固件，一切开源
git clone https://github.com/RobotBase/Q1-nano.git
cd Q1-nano详细文档请查阅 docs/ 目录。
欢迎任何形式的贡献！无论是修复文档、调参、训练新步态，都非常欢迎。
请先阅读 贡献指南。
If you find this project interesting, please give us a ⭐ — it helps a lot!
如果你觉得这个项目有意思，请给我们一个 ⭐ — 这对我们非常重要！