Microduck logo

Productivity / product dossier

Microduck

A 25 cm open-source biped for sim-to-real RL, with seven pretrained behaviors.

Product brief

What Microduck does.

Microduck is a 25cm, $399 open-source bipedal robot built by Hugging Face and Pollen Robotics. It’s designed for sim-to-real reinforcement learning. It ships with 7 pre-trained behaviors and an Apache 2.0 software stack you can clone, modify, and retrain.

Why we selected it

An unusually accessible open robotics platform for sim-to-real experimentation, especially timely amid growing releases of robot-training data.

Best for
Robotics researchers and RL builders
Category
Productivity
Daily picks
1
First selected
2026-08-28

Capability scan

What it can help with.

Only capabilities supported by the product information we collected are listed here.

01

Provide a bipedal robot platform

Offers a 25 cm open-source bipedal robot built by Hugging Face and Pollen Robotics.

02

Support sim-to-real reinforcement learning

Is designed for sim-to-real reinforcement-learning work.

03

Include pre-trained behaviors

Ships with seven pre-trained behaviors.

04

Enable software modification and retraining

Provides an Apache 2.0 software stack that users can clone, modify, and retrain.

Best-fit use cases

Experimenting with sim-to-real reinforcement learning.
Starting from one of the included pre-trained behaviors.
Modifying and retraining the provided robot software stack.
Using an open robotics platform for research or RL-building work.

FAQ

Before you open it.

What is Microduck?

Microduck is a 25 cm open-source bipedal robot from Hugging Face and Pollen Robotics.

What kind of learning is Microduck designed for?

It is designed for sim-to-real reinforcement learning.

Does Microduck include pre-trained behaviors?

Yes. It ships with seven pre-trained behaviors.

Can users modify its software?

Yes. Its Apache 2.0 software stack can be cloned, modified, and retrained.

Who is Microduck for?

It is intended for robotics researchers and reinforcement-learning builders.