datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
microduck-emotions
Microduck Emotions
A collection of emotions for the Microduck robot. Each one is a motion and a sound designed together, beat by
beat, with the beak opening on the sound, rendered in the physics simulation and validated on the real robot. Every
emotion is three files: the motion (emotions/<name>.json, keyframes at 30 fps: head and body offsets played on
top of whichever trained policy is active, plus the policy hand-overs, such as the sit that devastated and play dead
start)… See the full description on the dataset page: https://huggingface.co/datasets/pollen-robotics/microduck-emotions.microduck-electric-slide-motion
Microduck Electric Slide Motion
Current status — 2026-09-13: NOT a completed dance policy
The new 5 cm, exaggerated-step requirements have not passed. The 400-iteration phrase refinement stays upright in nominal GPU BAM evaluation, but fails step travel; one of ten randomized no-push rollouts falls, and the CPU BAM diagnostic also falls. No new training was run on September 13. Earlier statements below that all counts passed used weaker QC and do not establish… See the full description on the dataset page: https://huggingface.co/datasets/Histochemichael/microduck-electric-slide-motion.microduck-detection-dataset
Microduck Detection Dataset
Single-class object detection dataset (class 0 = microduck, the 25 cm biped robot duck announced by Pollen Robotics / Hugging Face on 2026-08-27) in YOLO format, built to train a tiny (<5M param) detector.
Composition
Synthetic scene renders (train ~1,040 / val ~60): the robot from the open-source MuJoCo model in pollen-robotics/microduck_rl (full visual-mesh model, scene_allcollisions.xml), rendered headlessly with randomized pose… See the full description on the dataset page: https://huggingface.co/datasets/pngwn/microduck-detection-dataset.microduck-skill-tree
Microduck skill tree, the training log
Simulation only. Nothing here has run on a real robot yet; the Microduck this is for is still on order. Every policy, curve and clip in this repo comes from mjlab / MuJoCo Warp on one RTX 5090, trained with the tasks in pollen-robotics/microduck_rl at commit 53b8971.
This is the build log. One folder per level of the skill tree, publishable or not, partial or not. The installable policies live in their own model repos (one per policy, in… See the full description on the dataset page: https://huggingface.co/datasets/witcheer/microduck-skill-tree.microduck-stairs-challenge
Microduck Stairs Challenge
Simulation only, never on hardware. Every number in this card was produced in MuJoCo,
in the plant sim/scene.mjb (sha256 3f8c9ab9b409ba74c73c30179d5f7c12b025f631693f9eec78d80dca242547be,
compiled from sim/scene_physics.xml, shipped here as harness/scene_physics.xml). No move
in this package has ever been run on a physical Microduck, and nothing here should be run on
one without a separate safety review: in the 20 cells where the trunk got over the 95… See the full description on the dataset page: https://huggingface.co/datasets/craigm26/microduck-stairs-challenge.microduck-hermes-runs
Microduck driven by Hermes Agent: simulation runs
One sentence typed to Hermes Agent, and a Pollen Robotics Microduck does it. Hermes Agent picks and runs every command itself. This dataset logs every such run: the sentence, the agent's full transcript, the simulator's ground truth, the robot daemon's own monitor, and a replay video.
Simulation only. Not yet tested on a real Microduck. The simulator runs Pollen's real robot daemon (robotd), so the commands, the safety layer and… See the full description on the dataset page: https://huggingface.co/datasets/witcheer/microduck-hermes-runs.microduck-trajectory-dataset
🦆 MicroDuck Bipedal Robot Simulation Trajectory Dataset
This dataset contains multi-modal state-action trajectories collected from the MicroDuck 14-DOF bipedal robot simulation in MuJoCo.
📂 Formats Available
.csv files: Human-readable tabular format containing timestamps, base pose (pos & quat), linear/angular velocities, 14 motor angles, velocities, applied torques (ctrl), and user velocity commands.
.npz files: Compressed NumPy tensor arrays (obs, actions… See the full description on the dataset page: https://huggingface.co/datasets/allen73/microduck-trajectory-dataset.microduck-locomotion-dataset-v1
🦆 MicroDuck Locomotion Dataset V1.0: Zero-Slip Flat Ground Baseline
Publisher: devorah-ai-2026 / devorahai2026License: Commercial / CC-BY-SA 4.0Task: Bipedal Locomotion, Reinforcement Learning, Sim-to-RealHardware Platform: Antigravity 14-DOF MicroDuck Biped Robot
📌 Overview (개요)
본 데이터셋은 **14-DOF 소형 2족 보행 로봇(MicroDuck)**의 완벽한 평면 보행(Flat-ground walking) 궤적 및 제어 로그를 담은 고품질 합성 데이터(Synthetic Data)입니다. 기존 강화학습 기반 2족 보행 에이전트들이 흔히 겪는 발바닥 미끄러짐(Ice-Skating) 및… See the full description on the dataset page: https://huggingface.co/datasets/devorah-ai-2026/microduck-locomotion-dataset-v1.microduck-locomotion-dataset-v1
🦆 MicroDuck Locomotion Dataset V1.0: Zero-Slip Flat Ground Baseline
Publisher: devorah-ai-2026 / devorahai2026License: Commercial / CC-BY-SA 4.0Task: Bipedal Locomotion, Reinforcement Learning, Sim-to-RealHardware Platform: Antigravity 14-DOF MicroDuck Biped Robot
📌 Overview (개요)
본 데이터셋은 **14-DOF 소형 2족 보행 로봇(MicroDuck)**의 완벽한 평면 보행(Flat-ground walking) 궤적 및 제어 로그를 담은 고품질 합성 데이터(Synthetic Data)입니다. 기존 강화학습 기반 2족 보행 에이전트들이 흔히 겪는 발바닥 미끄러짐(Ice-Skating) 및… See the full description on the dataset page: https://huggingface.co/datasets/devorah-ai-2016/microduck-locomotion-dataset-v1.microduck-ball-challenge
Microduck Ball Challenge
Simulation only, never on hardware. Every number in this card was produced in MuJoCo,
in the plant sim/scene.mjb (sha256 3f8c9ab9b409ba74c73c30179d5f7c12b025f631693f9eec78d80dca242547be,
compiled from sim/scene_physics.xml, shipped here as harness/scene_physics.xml;
results/chase_controls-results.json → plantDigest). Nothing in this package has been run on
a physical Microduck.
What the challenge is
The Pollen Robotics Microduck is a… See the full description on the dataset page: https://huggingface.co/datasets/craigm26/microduck-ball-challenge.microduck-jump-videos
Microduck jump demo videos
BAM actuator simulation vs PPO play recordings from local Microduck jump training (Mjlab-Jump-Flat-MicroDuck).
BAM (voltage/BAM actuator sim)
microduck_hop_jump_bam.mp4
microduck_jump_hop_bam.mp4
PPO play
microduck_hop_jump.mp4
microduck_jump_hop.mp4
microduck_walk_jump_hop.mp4
microduck_walk_jump_closeup.mp4
microduck_walk_jump_rise.mp4
jump_ppo_play_29999.mp4 (training play export @ iter 29999)
microduck-policy-golden-vectors
Microduck policy golden vectors
Observation → action pairs recorded from Pollen Robotics' trained
Microduck policies, so that anybody writing their
own runner can check it against the same numbers instead of against a video.
This is a conformance fixture, not a model and not a dataset to train on. It contains no
weights. If you want the networks, they are Pollen's, in
pollen-robotics/microduck and
pollen-robotics/microduck_rl.
What is in it
golden_policies.json… See the full description on the dataset page: https://huggingface.co/datasets/craigm26/microduck-policy-golden-vectors.microduck-feedback
Microduck community feedback
What people told Microduck Studio when they corrected a
proposed plan or chose between two walking ducks. One JSON object per line, in the
duck-feedback/0 format.
route_correction: a plain-language router proposed labels for one clause (for example
"turn left" → turn_left), and the person kept, edited or discarded them. These labels
recalibrate and fine-tune the router.
policy_preference: the person watched two trained walkers do the same thing in… See the full description on the dataset page: https://huggingface.co/datasets/craigm26/microduck-feedback.microduck-e2e-strands-robotsmicroduck-datasetmicroduck-video-artifacts