Hiwonder-robot/Cube2Plate
Robotics • 51.7M • Updated • 32
AI & ML interests
None defined yet.
Recent Activity
View all activity
Organization Card
This is the official Hiwonder hub on Hugging Face for open-source robotics, embodied AI, and robot learning resources.
Since the beginning, open source has been a core part of Hiwonder. We develop accessible hardware and software platforms for Embodied AI robotics research and education—from ROS robots and robotic arms to quadrupeds and humanoid robots.
Here you'll find models, datasets, demos, and tools designed to lower the barrier to entry for physical AI and embodied intelligence. Our goal is to help researchers, developers, educators, and makers build, train, and deploy intelligent robots more easily.
🧩 LeRobot-Compatible Featured Projects
Hiwonder SO-ARM101 and NexArm robot arms are natively integrated into the LeRobot ecosystem, supporting end-to-end imitation and reinforcement learning workflows. These arms leverage intuitive leader-follower teleoperation and a dual-camera setup, pairing an eye-in-hand gripper camera with an external global camera. This configuration seamlessly captures the rich Vision-Action datasets essential for training Vision-Language-Action (VLA) models.
Hardware lays the foundation, but its true potential is unlocked by what the community builds.
📦 With Hiwonder + LeRobot, You Can:
- Models: Pre-trained policies and fine-tuned robot learning models compatible with LeRobot and other modern frameworks.
- Datasets: Demonstration data captured via teleoperation for real-world manipulation tasks, ready for model training and performance benchmarking.
- Spaces: Interactive demos to visualize robot learning processes, explore datasets, and evaluate model performance.
- Collections: Dedicated resource bundles tailored for specific robotic platforms and research pipelines.
These resources are constantly growing as we expand support to more platforms, including our upcoming brushless servo systems.
🤖 Why Hiwonder
Purpose-Built Hardware
Developed with in-house high-precision, high-torque magnetic encoder servos, dual-vision systems, and durable mechanical designs, allowing developers to focus on algorithms instead of hardware debugging.
Comprehensive Open-Source Resources
Access open-source code, systematic tutorials, and comprehensive Wiki documentation, mapping out a seamless path from beginner to advanced mastery.
Global Community
Collaborate with researchers, developers, educators, and makers around the world who are building the future of embodied AI.
✨ How to Get Involved
We believe open collaboration drives robotics forward faster. You can contribute by:
- Share Datasets: Upload demonstration data captured on Hiwonder platforms.
- Contribute Models: Submit trained policies for the community to build upon.
- Build Spaces: Create interactive demos showcasing robot learning possibilities.
- Test & Validate: Run models on real hardware and share your benchmark feedback.
- Refine Documentation: Improve documentation, tutorials, and examples.
We deeply value every single contribution. For sustained contributors, we provide public acknowledgments, collaborator access, hardware support for testing, or other forms of community recognition. Building an open robotics ecosystem takes everyone's effort—let's shape the next generation together.