Hugging Face Datasets2026 · Table · Parquet
pigProfessional/so-arm101-stack-green_20260929_160851This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names": [ "shoulder_pan.pos", "shoulder_lift.pos", "elbow_flex.pos", "wrist_flex.pos", "wrist_roll.pos", "gripper.pos" ], "shape": [ 6… See the full description on the dataset page: https://huggingface.co/datasets/pigProfessional/so-arm10
Hugging Face Datasets2026 · dataset
MM-30MM-30 A real-world mobile manipulation dataset, released in a LeRobot v3.0 file layout. Episode snapshots Put fruit into a bowl Hang cups on a rack Sort blocks into cups Stack bowls Screw a cap onto a test tube Open a microwave Put fruit into a cabinet Load a refrigerator Fold a towel Fold clothing Open a bag and take out a toy Sweep blocks into a dustpan Overview This… See the full description on
Hugging Face Datasets2026 · Image
RawVLA-BenchRawVLA-Bench RawVLA-Bench is a paired training-trajectory dataset for studying RAW-domain visual frontends for vision-language-action (VLA) policies. It is derived from successful expert trajectories in LIBERO and RoboTwin 2.0. Each trajectory has a RAW input version captured or rendered under a sampled lighting condition and a paired default-light RGB target. The pair shares the same trajectory i
Hugging Face Datasets2026 · Table · Parquet
haijian06/yellow_cube_brown_box_v1This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names": [ "shoulder_pan.pos", "shoulder_lift.pos", "elbow_flex.pos", "wrist_flex.pos", "wrist_roll.pos", "gripper.pos" ], "shape": [ 6… See the full description on the dataset page: https://huggingface.co/datasets/haijian06/yellow_cube_br
Hugging Face Datasets2026 · Table · Parquet
Aaaay-0610/so101-task-1This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names": [ "shoulder_pan.pos", "shoulder_lift.pos", "elbow_flex.pos", "wrist_flex.pos", "wrist_roll.pos", "gripper.pos" ], "shape": [ 6… See the full description on the dataset page: https://huggingface.co/datasets/Aaaay-0610/so101-task-1.
Hugging Face Datasets2026 · Table · Parquet
kiteml/dual-openyam-close-boxThis dataset was created using LeRobot. Lighting augmentation The 61 original episodes were recorded in two lighting setups: episodes 0-30 in bright daylight, episodes 31-60 in dim, warm tungsten light. Each session was relit to look like the other one (same actions, states and task; only the camera pixels change, geometry and motion are untouched): Episodes 61-91: copies of 0-30 relit to dim tung
Hugging Face Datasets2026 · dataset
ssdunit/abc_sim_lego_blocks_sortingThis dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "observation.state": { "dtype": "float32", "shape": [ 14 ] }, "action": { "dtype": "float32", "shape": [ 14 ] }, "observation.images.top": { "dtype": "video", "shape":… See the full description on the dataset page: https://huggingface.co/datasets/ssdunit/abc_sim_lego_blo
Hugging Face Datasets2026 · Table · Parquet
kitare17/astra_pick_tThis dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "observation.state": { "dtype": "float32", "shape": [ 18 ], "names": [ "joint_l1", "joint_l2", "joint_l3", "joint_l4", "joint_l5", "joint_l6"… See the full description on the dataset page: https://huggingface.co/datasets/kitare17/astra_pick_t.
Hugging Face Datasets2026 · Table · Parquet
hungho77/so101-ai20k-multitaskThis dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names": [ "shoulder_pan.pos", "shoulder_lift.pos", "elbow_flex.pos", "wrist_flex.pos", "wrist_roll.pos", "gripper.pos" ], "shape": [ 6… See the full description on the dataset page: https://huggingface.co/datasets/hungho77/so101-ai20k-mul
Hugging Face Datasets2026 · Table · Parquet
Deviant65/so101_redcube_20260925_154203This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names": [ "shoulder_pan.pos", "shoulder_lift.pos", "elbow_flex.pos", "wrist_flex.pos", "wrist_roll.pos", "gripper.pos" ], "shape": [ 6… See the full description on the dataset page: https://huggingface.co/datasets/Deviant65/so101_redcube_
Hugging Face Datasets2026 · dataset
binhpham/reachy-mini-massive-motion-libraryReachy Mini Massive Motion Library 10,872 episodes (15.1 h at 25 fps). Each one is a prompt performed by Reachy Mini: the teacher's motion recipe, turned into 25 Hz motion by the plan-to-motion flow generator, projected onto the robot's reachable set, and rendered in MuJoCo with the official model. source episodes what astra 5,000 5,000 individually authored scenarios (Astra) astra_lively 5,000 th
Hugging Face Datasets2026 · dataset
Physical AI – UMI Bottle Cap Opening & Closing (Left Gripper, TRUMI)Gripper camera Left Gripper camera Right Physical AI – UMI Bottle Cap Opening & Closing (Left Gripper) A hand-held demonstration dataset of transferring objects from cups, recorded with the Trossen Robotics TRUMI gripper, an implementation of the Universal Manipulation Interface (UMI). The operator picks objects out of cups and transfers them to a new location using the TRUMI grippers. Raw GoPro v
Hugging Face Datasets2026 · dataset
Physical AI – UMI Bottle Cap Opening & Closing (Left Gripper, TRUMI)Gripper camera Left Gripper camera Right Physical AI – UMI Bottle Cap Opening & Closing (Left Gripper) A hand-held demonstration dataset of opening and closing a bottle cap, recorded with the Trossen Robotics TRUMI gripper, an implementation of the Universal Manipulation Interface (UMI). The left gripper performs the cap twisting motion. Raw GoPro video was processed with the TRUMI dataset generat
Hugging Face Datasets2026 · dataset
Physical AI – UMI Bottle Cap Opening & Closing (Left Gripper, TRUMI)Gripper camera Left Gripper camera Right Physical AI – UMI Bottle Cap Opening & Closing (Left Gripper) A hand-held demonstration dataset of opening and closing a bottle cap, recorded with the Trossen Robotics TRUMI gripper, an implementation of the Universal Manipulation Interface (UMI). The left gripper performs the cap twisting motion. Raw GoPro video was processed with the TRUMI dataset generat
Hugging Face Datasets2026 · dataset · gated
DexCanvasDexCanvas DexCanvas: Bridging Human Demonstrations and Robot Learning for Dexterous Manipulation Project page · Paper (arXiv:2510.15786) · Repository DexCanvas is a dataset of 111,857 complete right-hand trajectories and 55,442,683 frames of rigid hand–object interaction. Every Parquet row is one trajectory with a shared time axis for hand kinematics, object pose, and per-frame contacts. For each
Hugging Face Datasets2026 · Table · Parquet
witsense-ai/so101_longhorizon_datasetThis dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "observation.state": { "dtype": "float32", "shape": [ 6 ], "names": [ "joint_0", "joint_1", "joint_2", "joint_3", "joint_4", "joint_5" ] }… See the full description on the dataset page: https://huggingface.co/datasets/witsense-ai/so101_longhorizon_dataset.
Hugging Face Datasets2026 · Table · Parquet
alex07121/lerobot-so-arm-101This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names": [ "shoulder_pan.pos", "shoulder_lift.pos", "elbow_flex.pos", "wrist_flex.pos", "wrist_roll.pos", "gripper.pos" ], "shape": [ 6… See the full description on the dataset page: https://huggingface.co/datasets/alex07121/lerobot-so-arm
Hugging Face Datasets2026 · Table · Parquet
Aaaay-0610/so101-test-1_20260923_182449This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names": [ "shoulder_pan.pos", "shoulder_lift.pos", "elbow_flex.pos", "wrist_flex.pos", "wrist_roll.pos", "gripper.pos" ], "shape": [ 6… See the full description on the dataset page: https://huggingface.co/datasets/Aaaay-0610/so101-test-1_
Hugging Face Datasets2026 · Table · Parquet
Tridex/50_gazeuse_10h30_23-09_20260923_103845This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names": [ "shoulder_pan.pos", "shoulder_lift.pos", "elbow_flex.pos", "wrist_flex.pos", "wrist_roll.pos", "gripper.pos" ], "shape": [ 6… See the full description on the dataset page: https://huggingface.co/datasets/Tridex/50_gazeuse_10h30_
Hugging Face Datasets2026 · Image
EmbodiedMemory-BenchEmbodiedMemory-Bench Benchmarking Embodied Memory for Long-Horizon Embodied Tasks Remember what the world was. Track how it changed. Act beyond the moment. 2,554 episodes · 4 memory challenges · AI2-THOR & ProcTHOR Benchmark EmbodiedMemory-Bench (EMem-Bench) evaluates memory through actions in an interactive environment. An agent first observes multimodal interaction history without knowing its fu
Hugging Face Datasets2026 · dataset
AutoLab-SJTU/robofollow-dataRoboFollow: Unveiling the Instruction Following Mirage in Embodied Agents This dataset is part of the RoboFollow benchmark, designed to evaluate instruction-following capabilities of embodied agents in manipulation tasks. Built on RoboTwin, it provides shared-scene task configurations where multiple tasks coexist, requiring policies to rely on language to disambiguate objectives. The dataset inclu
Hugging Face Datasets2026 · Table · Parquet
EgoSteer Egocentric Human Hand-Motion DatasetEgoSteer Egocentric Human Hand-Motion Dataset Overview EgoSteer-Egocentric contains 1,544,579 episodes of bimanual human hand motion, from 8 public egocentric video datasets: estimated from the video by the EgoSmith pipeline, or converted from a source's official annotation where it has one (see the Source Datasets table). Every frame carries metric 3D wrist poses and fingertip positions of both h
Hugging Face Datasets2026 · Text
ChangUoA/FailTimeFailTime Failure annotations for robot manipulation rollouts, from the paper ProTracer: Proprioception-Guided Failure Diagnosis in Robot Manipulation. Each episode is labelled as a success or a failure, and every failure carries the moment it went wrong, its failure type, the reason, and how it could have been avoided. FailTime-Short — annotations for ViFailback recordings (AgileX Piper). Download
Hugging Face Datasets2026 · Table · Parquet
BlackWhite123/fold_clothThis dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names": [ "shoulder_pan.pos", "shoulder_lift.pos", "elbow_flex.pos", "wrist_flex.pos", "wrist_roll.pos", "gripper.pos" ], "shape": [ 6… See the full description on the dataset page: https://huggingface.co/datasets/BlackWhite123/fold_cloth
Hugging Face Datasets2026 · Table · Parquet
HOI-Retarget — Humanoid Human-Object Interaction MotionHOI-Retarget — Humanoid Human–Object Interaction Motion Project page · Code · 3D viewer · Paper · arXiv Learning from demonstration (LfD) has enabled humanoid robots to acquire diverse whole-body skills, but extending this paradigm to human-object interaction (HOI) is limited by the availability of robot-compatible interaction references. We present HOI-Retarget, a contact-centric retargeting meth