Hardware verifiedrecoveryv1.0.0Apache-2.0

Dynamic Fall Recovery

Automatic fall recovery policy capable of restoring MicroDuck to standing posture from face-down, face-up, or sitting positions.

By Pollen Robotics @pollen-robotics

MicroDuck pushed over on carpet recovering to standing posture

About this behavior

Triggered automatically when the onboard IMU's projected gravity vector exceeds tilt thresholds (or commanded via gamepad recovery button). The policy executes robust arm-less biped pushups, righting the duck without human intervention before handing off control back to the standing/walking state machine.

Runtime contract

61-dimension input · 14 joint targets · 50 Hz

Show details

50 Hz / 4×slot: stand

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 fall detection triggers enabled. Robot self-rights after tip-over.
Target hardware: MicroDuck v1 (Rockchip RK3566, 14 Dynamixel XL330-M077, 2S LiPo)
Simulation: mjlab (MuJoCo Warp) at 50 Hz
Note: Requires full body contact physics. Uses robot_allcollisions.xml in simulation.

Compatibility

Robot model: microduck-standard
MJCF model: robot_allcollisions.xml
Terrain: flatrough
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: stand

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_stand.onnx

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

2 Add the policy to robotd

[policy]
stand = "/opt/robot/policies/BEST_alpha_stand.onnx"

3 Restart the control loop

sudo systemctl restart robotd
robotctl health

⌁ Try it in simulation

uv run scripts/infer_policy.py --standing BEST_alpha_stand.onnx --new-cmd-obs

Sources

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