Hugging Face has unveiled 'Microduck', a $399 open-source, duck-shaped robot designed to democratize physical AI. Powered by reinforcement learning and equipped with advanced sensors, this developer-friendly bot can perform complex tasks like picking up items and even roller skating.
- Hugging Face has launched 'Microduck,' an affordable $399 open-source robot designed for physical AI development.
- The 25-centimeter-tall robot features advanced sensors, including LiDAR and IMUs, and can perform tasks like lifting up to 800 grams and roller skating.
- This launch comes amidst reports of a pending $13 billion acquisition of Hugging Face by tech giant Nvidia.
AI powerhouse Hugging Face has officially stepped into the physical world with the unveiling of the Microduck. Priced at a highly accessible $399, this cute, duck-like robot is designed for developers to train right out of the box using reinforcement learning. The company has confirmed that pre-orders are open, with shipments scheduled to arrive before Christmas.
According to Clem Delangue, CEO of Hugging Face, the Microduck represents a milestone in making physical AI accessible to everyone. Standing at just 25 centimeters tall, this miniature robot possesses an impressive array of physical capabilities. It can waddle, bend, recover its balance after falling, pick up objects weighing up to 800 grams with its beak, and even roller skate on flat surfaces.
"Welcome to the era of open-source affordable robots to democratize physical AI and world models!" – Clem Delangue, CEO of Hugging Face
Advanced Hardware and Seamless Simulation
Under the hood, the Microduck is packed with sophisticated spatial awareness technology. It perceives its environment using a high-definition camera, LiDAR sensors, and two Inertial Measurement Units (IMUs) that track orientation and motion. Developed in collaboration with French startup Pollen Robotics, the robot leverages a seamless workflow where behaviors can be trained in a virtual simulation and deployed directly onto the physical hardware.
Developers can access the complete Software Development Kit (SDK), simulation environments, and the reinforcement learning (RL) training stack directly on GitHub. This open ecosystem allows researchers to fine-tune, retrain, and redeploy models dynamically, bypassing the proprietary restrictions typical of industrial robotics.
Historical Background: Hugging Face’s Hardware Journey
While Hugging Face is globally renowned as the leading platform for sharing open-source AI model weights, its foray into hardware began in earnest in April 2025 with the acquisition of Pollen Robotics. Since then, the partnership has yielded affordable desktop robots like the Reachy Mini ($499) and the Reachy Mini Lite ($399). The Microduck is the latest evolution of this initiative, shifting the focus toward ultra-portable, highly interactive physical agents.
Why This Matters
BozokMedia analysis shows that the release of the Microduck is a calculated move to challenge the dominance of proprietary tech conglomerates in the robotics sector. By lowering the financial barrier to entry, Hugging Face is fostering an ecosystem where independent developers and academic institutions can contribute to physical AI research without needing multi-million dollar laboratory setups.
The table below provides a comparison of Hugging Face's current hardware lineup:
| Feature | Microduck | Reachy Mini | Reachy Mini Lite |
|---|---|---|---|
| Price | $399 | $499 | $399 |
| Processor Support | Simulation / On-board SDK | Raspberry Pi | Mac or PC |
| Sensors | Camera, LiDAR, 2x IMUs | Camera, Standard Sensors | Standard Sensors |
| Unique Ability | Roller skating, 800g lift capacity | Desktop assistance | Basic manipulation |
Privacy, Security, and Corporate Speculation
As with any home-deployed device equipped with cameras and microphones, privacy remains a paramount concern. Clem Delangue has previously defended open-source robotics, arguing that auditability makes them inherently safer than "black box" proprietary systems controlled by centralized corporations. However, security experts warn that the ultimate responsibility lies with the users, as third-party applications built on top of the open SDK could potentially transmit sensor data to external servers.
This launch is further amplified by industry rumors that Nvidia is preparing to acquire Hugging Face in a deal valued at $13 billion. Nvidia has been a key infrastructure provider for the startup since 2023, and both companies share a public commitment to advancing open-source AI architectures.
Frequently Asked Questions
Q1: Is the Microduck suitable for non-programmers?
A1: While the Microduck is highly adorable, it is primarily a development kit. Utilizing its full capabilities requires familiarity with Python, reinforcement learning, and basic robotics programming.
Q2: Where can I access the code to train the Microduck?
A2: The full SDK, simulation tools, and reinforcement learning training stack are publicly available on Hugging Face's official GitHub repository.
