Microduck
A 25 cm open-source biped robot trained with reinforcement learning
Visit pollen-robotics.com ↗What is Microduck?
Microduck is a compact biped robot designed for learning and experimentation. It features 15 motors, a camera, LiDAR, and a grasping beak, arriving ready to use out of the box. The robot's open-source stack enables users to train new behaviors in simulation using reinforcement learning, then deploy them directly to the hardware. The project indicates it was built with Claude AI coding tools.
Users can teach the robot various skills—walking, sitting, kicking, grabbing objects, and recovering from falls—by retraining policies on their own machines or via Hugging Face Jobs. Each trained behavior can be shared with the community. The sim-to-real workflow lets developers iterate quickly between simulation and physical deployment, making it accessible for robotics experimentation and education.
Momentum
Daily activity over the last 30 days: upvotes, unique visits, new GitHub stars and Hacker News points.
How do we know it's vibe coded?
vibetrend discovered Microduck automatically. Evidence: GitHub pollen-robotics/microduck; repo has AGENTS.md, CLAUDE.md. View source.
Similar vibe-coded apps
#6 this week523 free lessons building AI systems from math and first principles

Project-based course on building reliable AI coding agent environments

Learn AI coding by building real products, from zero to shipped

Visualize agent loops with local LLM inference and token-level transparency


C. elegans brain simulation steering a language model in your browser

Interactive climate simulation built from first-principles physics

Interactive browser visualization of LLM agent loops and workflows