Microduck MD-01 combines ready-to-run play with a real development platform. It can walk, crouch, grasp with its articulated beak, recover from a fall and use optional rollers. Seven published policies are supplied as a starting point, and the open software workflow lets you iterate in simulation before testing on the physical robot.
Ready on day one
Use the included gamepad and built-in behaviours before opening the development tools.
Perception on board
A wide-angle camera, compact 8x8 ToF LiDAR, two IMUs, microphones, a speaker and two NFC antennas support interactive projects.
Built for physical AI
The RK3566 computer, AI accelerator and 50 Hz onboard policy loop bring trained behaviours onto the robot.
Open development path
Create and tune reinforcement-learning policies in MuJoCo, deploy conservatively, observe and refine.
A strong fit for
- Developers exploring reinforcement learning and sim-to-real workflows
- Educators demonstrating motion, perception and embodied AI
- Makers who want a playful robot they can continue to program
- Supervised demonstrations and interactive experiments