Hardware verifiedlocomotionv1.0.0Apache-2.0

Alpha Dynamic Walk

Standard bipedal walking gait with 50 Hz velocity tracking (vx, vy, yaw_rate) and active head-pose orientation control.

MicroDuck walking under gamepad velocity control on hardware

About this behavior

The flagship locomotion policy for MicroDuck. Trained in MuJoCo Warp via mjlab and PPO using the BAM M6 actuator model for Dynamixel XL330 servos. Includes domain randomization on battery voltage, friction, and motor command delay. Exports a baked observation normalizer into the ONNX computational graph.

Runtime contract

61-dimension input · 14 joint targets · 50 Hz

Show details

50 Hz / 4×slot: walk

Observation vector61 dimensions · normalizer baked in

Proprioception48D

Positions, velocities, previous actions, gravity, and gyro.

Twist command3D

Forward, lateral, and yaw velocity intent.

Head pose4D

Neck and head orientation targets.

Body pose6D

Six-axis torso offset command.

Action targetsscale: 1×

Left leg5 joints

hip yaw · hip roll · hip pitch · knee · ankle

Neck & head4 joints

neck pitch · head pitch · yaw · roll

Right leg5 joints

hip yaw · hip roll · hip pitch · knee · ankle

Actuator model: Dynamixel XL330 (BAM M6 voltage control law). The beak motor is separate from this fixed locomotion action space.

Verification

Status: Hardware verified
Evidence: Verified on physical MicroDuck hardware with Rockchip RK3566 and 14 Dynamixel XL330 servos. Shipped as default walking policy.
Target hardware: MicroDuck v1 (Rockchip RK3566, 14 Dynamixel XL330-M077, 2S LiPo)
Simulation: mjlab (MuJoCo Warp) at 50 Hz
Note: Operates at 50 Hz control loop. Velocity envelope: -0.2 to +0.25 m/s forward/back, ±1.0 rad/s angular yaw rate.

Compatibility

Robot model: microduck-standard
MJCF model: robot_walk.xml
Terrain: flat
Required accessories: None — standard duck setup

Run it on your duck

Use the canonical model and this policy slot to try the behavior locally.

slot: walk

Try the simulation first. Hardware testing assumes a matching Microduck, clear space, and a safe surface. uDuck does not guarantee that a community policy is safe to run.

Deployment stepsShow instructions

1 Download the ONNX model BEST_alpha_walking.onnx

curl -L "https://huggingface.co/spaces/pollen-robotics/microduck-simulator/resolve/main/app/public/policies/BEST_alpha_walking.onnx" -o "/opt/robot/policies/BEST_alpha_walking.onnx"

2 Add the policy to robotd

[policy]
walk = "/opt/robot/policies/BEST_alpha_walking.onnx"

3 Restart the control loop

sudo systemctl restart robotd
robotctl health

⌁ Try it in simulation

uv run scripts/infer_policy.py --walking BEST_alpha_walking.onnx --new-cmd-obs

Sources

Upstream repositorypollen-robotics/microduck Training codeMjlab-Velocity-Flat-MicroDuck Interactive simulatorhttps://huggingface.co/spaces/pollen-robotics/microduck-simulator