{
  "schema_version": 1,
  "generated_at": "2026-07-25T05:20:00Z",
  "datasets": [
    {
      "slug": "droid-1-0-1",
      "display_name": "DROID 1.0.1",
      "summary": "A large in-the-wild robot manipulation corpus collected across diverse scenes and hardware.",
      "category": "generalist",
      "tags": ["robotics", "manipulation", "real-world", "video", "lerobot"],
      "source": {
        "provider": "huggingface",
        "repo_id": "lerobot/droid_1.0.1",
        "revision": "0eabc778f959c54b8c5aa3626cc1128d2d2e54d4",
        "gated": false,
        "license": "apache-2.0",
        "license_url": "https://huggingface.co/datasets/lerobot/droid_1.0.1"
      },
      "format": "LeRobot / Parquet",
      "modalities": ["video", "state", "action", "language"],
      "size_label": "10M-100M rows",
      "storage_gb": 809.9,
      "robot": "Multi-robot",
      "tasks": ["in-the-wild manipulation", "imitation learning"],
      "curation": {"status": "curated", "notes": "Large download; use filtered or streaming workflows."}
    },
    {
      "slug": "modified-libero-rlds",
      "display_name": "Modified LIBERO RLDS",
      "summary": "The OpenVLA-ready LIBERO benchmark conversion for language-conditioned manipulation research.",
      "category": "simulation",
      "tags": ["robotics", "manipulation", "libero", "rlds", "benchmark"],
      "source": {
        "provider": "huggingface",
        "repo_id": "openvla/modified_libero_rlds",
        "revision": "6ce6aaaaabdbe590b1eef5cd29c0d33f14a08551",
        "gated": false,
        "license": "mit",
        "license_url": "https://huggingface.co/datasets/openvla/modified_libero_rlds"
      },
      "format": "RLDS / TFRecord",
      "modalities": ["image", "state", "action", "language"],
      "size_label": "10.2 GB",
      "storage_gb": 10.2,
      "robot": "LIBERO simulated arm",
      "tasks": ["spatial", "object", "goal", "long-horizon"],
      "curation": {"status": "curated", "notes": "Prepared for OpenVLA LIBERO evaluation workflows."}
    },
    {
      "slug": "pusht",
      "display_name": "PushT",
      "summary": "A compact planar pushing benchmark widely used for diffusion-policy and imitation-learning baselines.",
      "category": "simulation",
      "tags": ["robotics", "pushing", "diffusion-policy", "video", "lerobot"],
      "source": {
        "provider": "huggingface",
        "repo_id": "lerobot/pusht",
        "revision": "7628202a2180972f291ba1bc6723834921e72c19",
        "gated": false,
        "license": "mit",
        "license_url": "https://huggingface.co/datasets/lerobot/pusht"
      },
      "format": "LeRobot / Parquet",
      "modalities": ["video", "state", "action"],
      "size_label": "10K-100K rows",
      "storage_gb": 0.19,
      "robot": "Planar PushT agent",
      "tasks": ["object pushing", "trajectory prediction"],
      "curation": {"status": "curated", "notes": "Small benchmark suitable for local experimentation."}
    },
    {
      "slug": "aloha-transfer-cube-human",
      "display_name": "ALOHA Transfer Cube Human",
      "summary": "Human teleoperation demonstrations for the bimanual ALOHA cube-transfer task.",
      "category": "manipulation",
      "tags": ["robotics", "bimanual", "aloha", "teleoperation", "lerobot"],
      "source": {
        "provider": "huggingface",
        "repo_id": "lerobot/aloha_sim_transfer_cube_human",
        "revision": "6a43d500f101255823a9d2b9dc244eeb01a2cd31",
        "gated": false,
        "license": "mit",
        "license_url": "https://huggingface.co/datasets/lerobot/aloha_sim_transfer_cube_human"
      },
      "format": "LeRobot / Parquet",
      "modalities": ["video", "state", "action"],
      "size_label": "10K-100K rows",
      "storage_gb": 3.12,
      "robot": "ALOHA bimanual",
      "tasks": ["cube transfer", "bimanual manipulation"],
      "curation": {"status": "curated", "notes": "Human demonstration split for ACT-style policy experiments."}
    },
    {
      "slug": "metaworld-mt50",
      "display_name": "MetaWorld MT50",
      "summary": "A 50-task manipulation benchmark packaged for multi-task policy training and evaluation.",
      "category": "simulation",
      "tags": ["robotics", "manipulation", "multi-task", "metaworld", "lerobot"],
      "source": {
        "provider": "huggingface",
        "repo_id": "lerobot/metaworld_mt50",
        "revision": "a59f742d218c903328164257ecf180f9b18018a1",
        "gated": false,
        "license": "apache-2.0",
        "license_url": "https://huggingface.co/datasets/lerobot/metaworld_mt50"
      },
      "format": "LeRobot / Parquet",
      "modalities": ["image", "state", "action"],
      "size_label": "100K-1M rows",
      "storage_gb": 91.75,
      "robot": "MetaWorld Sawyer",
      "tasks": ["50 manipulation tasks", "multi-task learning"],
      "curation": {"status": "curated", "notes": "Broad simulated task coverage; substantial local storage required."}
    },
    {
      "slug": "xarm-lift-medium",
      "display_name": "XArm Lift Medium",
      "summary": "A compact XArm lifting dataset for behavior cloning and offline reinforcement-learning baselines.",
      "category": "manipulation",
      "tags": ["robotics", "xarm", "lifting", "offline-rl", "lerobot"],
      "source": {
        "provider": "huggingface",
        "repo_id": "lerobot/xarm_lift_medium",
        "revision": "79efb0e3cef0e530ddec4b8569b190966ab45808",
        "gated": false,
        "license": "mit",
        "license_url": "https://huggingface.co/datasets/lerobot/xarm_lift_medium"
      },
      "format": "LeRobot / Parquet",
      "modalities": ["video", "state", "action"],
      "size_label": "10K-100K rows",
      "storage_gb": 0.35,
      "robot": "XArm",
      "tasks": ["object lifting", "offline policy learning"],
      "curation": {"status": "curated", "notes": "Compact dataset suitable for teaching and local baselines."}
    }
  ]
}