{
  "schema_version": 1,
  "name": "Actionable AI data catalogue",
  "url": "https://actionable.ai/#catalogue",
  "access": "Sample access, licensing and delivery are confirmed by enquiry.",
  "contact": "info@actionable.ai",
  "collections": [
    {
      "id": "E01",
      "name": "Stereo Ego · 1080p",
      "url": "https://actionable.ai/datasets/E01.html",
      "category": "Egocentric",
      "description": "Human perspective, captured in stereo with audio and motion sensing.",
      "inventory": "4,000 hours",
      "sample_recordings": 17,
      "capture_system": "Head-mounted stereo camera",
      "resolution": "2 × 1920 × 1080",
      "frame_rate": "30 fps",
      "format": "H.265 / MCAP",
      "views": "2 video streams",
      "package": "Each sample includes an MCAP recording and a matching undistorted MP4 preview.",
      "coverage": "17 distinct sample tasks across 7 scene domains.",
      "annotation": "Stereo inventory: 615 environment categories and 9,161 task types, with clip-level annotation.",
      "tasks": [
        "Vehicle engine-bay maintenance",
        "Pet bathing",
        "Preparing skewers",
        "Pharmacy stocktaking",
        "Retail shelf replenishment",
        "Garment finishing",
        "Metal polishing",
        "Cooking and plating",
        "Folding clothes",
        "Office document handling",
        "Floor and window cleaning"
      ]
    },
    {
      "id": "E02",
      "name": "Stereo Ego · Wide-view",
      "url": "https://actionable.ai/datasets/E02.html",
      "category": "Egocentric",
      "description": "A wider perspective on complete manufacturing and craft workflows.",
      "inventory": "70,000 hours",
      "sample_recordings": 6,
      "capture_system": "CyberCap2 head-mounted stereo camera",
      "resolution": "2 × 1920 × 1200",
      "frame_rate": "30 fps",
      "format": "H.265 / MCAP",
      "views": "2 video streams",
      "package": "Each sample includes an MCAP recording and a matching calibration JSON. Clips run approximately 5–7 minutes.",
      "coverage": "Manufacturing and craft workflows across six sample tasks.",
      "annotation": "Stereo inventory: 615 environment categories and 9,161 task types, with clip-level annotation.",
      "tasks": [
        "Circuit-board soldering",
        "Paper-bag assembly",
        "Small-part insertion",
        "Pen assembly",
        "Balloon inflation",
        "Textile wrapping and packing"
      ]
    },
    {
      "id": "E03",
      "name": "Monocular Ego · Android",
      "url": "https://actionable.ai/datasets/E03.html",
      "category": "Egocentric",
      "description": "Everyday skills and workplace tasks, seen from a human point of view.",
      "inventory": "40,000 hours",
      "sample_recordings": 18,
      "capture_system": "Head-mounted Android phone",
      "resolution": "1920 × 1080 + portrait formats",
      "frame_rate": "30 fps",
      "format": "H.264 / H.265 · MP4",
      "views": "1 video stream",
      "package": "MP4 clips organized by task category; two examples in each of nine sample categories.",
      "coverage": "8 top-level inventory scene categories; 9 task categories in the sample pack.",
      "annotation": "8 top-level inventory scene categories.",
      "tasks": [
        "Office and document work",
        "Packing and unpacking",
        "Selection and sorting",
        "Cleaning",
        "Shelf replenishment",
        "Repair and maintenance",
        "Assembly and production",
        "Food preparation",
        "Food service"
      ]
    },
    {
      "id": "E04",
      "name": "Monocular Ego · iPhone",
      "url": "https://actionable.ai/datasets/E04.html",
      "category": "Egocentric",
      "description": "Continuous first-person recordings that follow activities from start to finish.",
      "inventory": "10,000 hours",
      "sample_recordings": 13,
      "capture_system": "Head-mounted iPhone",
      "resolution": "1920 × 1080 / 1920 × 1440",
      "frame_rate": "30 / 60 fps",
      "format": "H.265 · MP4 / MOV",
      "views": "1 video stream",
      "package": "MP4 and MOV recordings without fixed-length slicing, including longer continuous task cycles.",
      "coverage": "11 inventory scene categories and 9 task categories; 9 scene domains in the sample pack.",
      "annotation": "Cross-annotation across 11 scene categories and 9 task categories.",
      "tasks": [
        "Retail stocking",
        "Food preparation",
        "Public-area cleaning",
        "Washing household items",
        "Metal forming",
        "Desk organization",
        "Fruit selection",
        "Toy assembly",
        "Note-taking",
        "Flower preparation",
        "Computer cabling"
      ]
    },
    {
      "id": "U01",
      "name": "Bimanual UMI",
      "url": "https://actionable.ai/datasets/U01.html",
      "category": "UMI",
      "description": "Connect human demonstrations to precise gripper movement and action.",
      "inventory": "10,000+ episodes",
      "sample_recordings": 10,
      "capture_system": "Two handheld grippers + head-mounted camera",
      "resolution": "3 × 1600 × 1200",
      "frame_rate": "60 fps",
      "format": "H.264 · MP4 + MCAP",
      "views": "3 video streams",
      "package": "Three video views, pose/action sequences, camera calibration and a trajectory-projection preview per episode.",
      "coverage": "Household and office manipulation across ten sample tasks.",
      "annotation": "Task-level labels; can be reorganized to a customer-defined skill taxonomy.",
      "tasks": [
        "Placing fruit in a bowl",
        "Organizing condiment bottles",
        "Clearing dishes",
        "Setting tableware",
        "Sorting laundry",
        "Collecting toys",
        "Storing household items",
        "Organizing computer peripherals",
        "Sorting snacks",
        "Folding clothes"
      ]
    },
    {
      "id": "R01",
      "name": "Real Robot · G1 / G2",
      "url": "https://actionable.ai/datasets/R01.html",
      "category": "Robot",
      "description": "Whole-body and dual-arm experience from robots in real environments.",
      "inventory": "20,000 hours",
      "sample_recordings": 53,
      "capture_system": "AGIBOT Genie-01 / Genie-02 + OmniPicker grippers",
      "resolution": "8–9 cameras, up to 1920 × 1536",
      "frame_rate": "30 fps; G2 home stereo 60 fps",
      "format": "LeRobot v3.0",
      "views": "8–9 video streams",
      "package": "One independent LeRobot v3.0 dataset per task, organized by robot, scene, location and task.",
      "coverage": "53 episodes across 11 robot/task groups in homes, a hotel and a library.",
      "annotation": "Scene → location → task hierarchy, with an individual task description for each recording.",
      "tasks": [
        "Sofa tidying",
        "Tabletop rubbish removal",
        "Sandwich preparation",
        "T-shirt folding",
        "Hotel changing-area organization",
        "Library shelf organization",
        "Packing stationery"
      ]
    },
    {
      "id": "R02",
      "name": "Galbot Robot Data",
      "url": "https://actionable.ai/datasets/R02.html",
      "category": "Robot",
      "description": "Multi-view robot demonstrations across household manipulation tasks.",
      "inventory": "20,000 QC-passed hours",
      "sample_recordings": 17,
      "capture_system": "Galbot robot",
      "resolution": "Head: 1280 × 960; hand: 640 × 360",
      "frame_rate": "30 fps",
      "format": "RGB video + MCAP",
      "views": "4 RGB views",
      "package": "Multi-view video, MCAP, joint data and odometry. Audio is available in 16 of the 17 listed sample packages.",
      "coverage": "13 task types across 17 sample recordings.",
      "annotation": "9 task categories and 6 scene categories.",
      "tasks": [
        "Place the floor mat in the designated position",
        "Place the drinking cup on the coaster",
        "Fold a blanket and place it on the sofa",
        "Place pillows at the head of the bed and align them",
        "Stack dinner plates",
        "Place water bottles and close the refrigerator door",
        "Smooth the bedsheet and align it with the bed edges",
        "Nest mixing bowls and close the cabinet door",
        "Push the storage box under the bed and align it",
        "Return the storage box to the drawer and close the drawer firmly",
        "Organize bedding and keep it tidy",
        "Fold and organize kitchen cloths",
        "Squeeze sauce onto a plate"
      ]
    },
    {
      "id": "R03",
      "name": "Long-Horizon Robot Data",
      "url": "https://actionable.ai/datasets/R03.html",
      "category": "Robot",
      "description": "Multi-step, bimanual demonstrations for learning sequences of action.",
      "inventory": "Inventory volume available on enquiry",
      "sample_recordings": 12,
      "capture_system": "Yuanli Lingji / dos_w1",
      "resolution": "640 × 480 RGB",
      "frame_rate": "30 fps",
      "format": "RGB MP4 + depth MKV + action JSONL",
      "views": "3 RGB + 3 depth views",
      "package": "RGB video, depth video and action sequences for multi-step and bimanual manipulation. Arm actions at 250 Hz; mobile-base actions at 50 Hz.",
      "coverage": "4 long-horizon and 8 manipulation sample recordings.",
      "annotation": "Descriptive task sequences; per-step timing is not included.",
      "tasks": [
        "Retrieve a power bank from a drawer and pack it into a backpack",
        "Pack stationery into a pencil case and return it to a drawer",
        "Stack blocks",
        "Pour particles from a cup into a box",
        "Bimanual bowl handover and shelf placement"
      ]
    }
  ]
}