@mattshumer_: This is so damn cool! And by the time it ships, models will be insanely good at setting up RL runs to train this autono…
Summary
Thomas Wolf unveils Microduck, the first truly accessible open-source biped robot designed for reinforcement learning training.
View Cached Full Text
Cached at: 08/28/26, 09:55 AM
This is so damn cool!
And by the time it ships, models will be insanely good at setting up RL runs to train this autonomously.
@huggingface @pollenrobotics would love to try it early and share my thoughts/experiments with the world!
Thomas Wolf (@Thom_Wolf): We have a huge news to share today!
Today we are unveiling the first truly accessible RL robot - welcome Microduck
A 25 cm tiny open-source biped with 15 actuators and packed with sensors (camera, speaker, LiDAR, NFC, bluetooth, wifi, etc) that you train yourself with
Similar Articles
Thousands of people are about to start training behaviors on real, tiny humanoid robots
Pollen Robotics releases Microduck, an open-source biped humanoid robot with a sim2real pipeline, enabling enthusiasts to train behaviors using reinforcement learning and potentially accelerating innovation in robotics.
Looks like we are close to local llama robotics
Pollen Robotics unveils Microduck, an open-source biped robot trainable with reinforcement learning, priced under $400, featuring pre-trained policies and a full open training stack.
@victormustar: you don't even need to own Microduck to start having fun with it - the simulator works great you can just ask your agen…
Pollen Robotics showcases Microduck, a robot and simulator with an open-source reinforcement learning stack, enabling sim2real training for AI agents.
@yacineMTB: they shipped an open source simulator for it i bought this immediately
Pollen Robotics announced Microduck, a small biped robot with an open source simulator, priced at $399 and shipping before Christmas.
@pham_blnh: just had a quick look at microduck_rl the codebase is very elegant, i’d recommend reading the agent.md since it contain…
A GitHub repository providing RL training environments for the Microduck bipedal robot, built on MuJoCo Warp with PPO, including sim2real deployment and reward modeling techniques.