A 25 cm open-source biped you train yourself with reinforcement learning. Playable out of the box.

The launch film · sound on


Waking up the duck…Meet the twin
Trained in sim, deployed on the real robot. This is the simulated twin the ducks were trained on.
Launch the simulator
Fun out of the box. Yours to retrain.
Every behaviour is a policy you can retrain on your own machine.
Behaviours are learned in physics sim, on your machine or on Hugging Face Jobs.
One step from simulation to the real thing.
Tune, re-train, re-deploy.
Share your new behavior with the community!
Velocity-tracking gait.
Sits down, holds the pose, stands back up on its own.
A one-shot boot, then straight back to walking.
Dips the beak to the ground, scoops, and pops back upright.
Roller skating locomotion when the skates are equipped.
Flat on its back to standing, all by itself, ready for the next command.
One robot, four colourways
Every Microduck ships in one of four colourways. Same robot, same brains underneath — pick the shell that best fits you.
Waking up the duck…Out in the world
The real robot in real places — on desks, on the pitch, out at golden hour.

View
View
View
View
View
View
View
View
View
View
View
View

ViewThe robot, and what to add to it
The robot is everything you need on day one. The packs add play gear and spare parts.

Built in the open
The SDK, the simulation and the full RL training stack are on GitHub. What the robot runs is what you can read, fork and retrain.
The whole software stack, permissively licensed. The physics sim every policy is trained in. Every shipped move, published and retrainable.
pollen-robotics/microduck
Builds on show, policies to swap, help when a leg does something strange. The community lives on Discord.
Join the DiscordEnd of tape · be kind, rewind
In four colourways. Introductory price, before taxes and shipping.
Pre-order for $399