craigm26/microduck-duckbatch-social

duckbatch-social

Four ducks who have never met, each run by its own brain: a small behavior engine with states named after Pollen's docs/ideas/autonomous_behavior.md (Chill, LookAround, Wander, TurnInPlace, Greet, Dance, Nap). The brains ran closed loop inside the Arena's simulator, sensing every duck's true pose ten times a second, and this timeline is the record of what they chose. Three walkers are duckbatch's b003e (Book 4 at microduckstudio.com); the napping duck is Pollen's got-a-talent reference. Greetings play the curious and yes head moves from pollen-robotics/microduck-emotions; dances bob on a shared half-second beat. The same brain will drive live ducks next.

A perpetual policy for the microduck (61-D observation, 14 actions, 50 Hz). Runs until told otherwise — a gait for the walk slot.

Run it on a robot

sudo robotctl policy load walk craigm26/microduck-duckbatch-social

The observation normalizer is baked into policy.onnx; feed raw observations. manifest.json follows schema 2 of the microduck policy manifest (docs/policy-manifest.md in the daemon repo).

A remix of craigm26/microduck-duckbatch-b002-128x128.

Training

• repo: pollen-robotics/microduck_rl