{
  "slug": "idb-multimodal-robot-cable-manipulation-benchmark-714021",
  "title": "IDB: Multimodal Robot Cable Manipulation Benchmark",
  "dek": "A concise explainer of IDB, DAG-ROS, and AG-iDP3: benchmark boards and multimodal imitation learning for industrial dexterous manipulation, with a reported 78% cable-task success rate.",
  "summary": "IDB benchmark explained: boards, DAG-ROS, AG-iDP3, and a 78% vs 36% robot cable-manipulation result.",
  "tags": [
    "robotics",
    "dexterous-manipulation",
    "imitation-learning",
    "diffusion-policy",
    "industrial-automation",
    "robot-benchmark",
    "cable-manipulation",
    "multimodal-learning"
  ],
  "published_at": "2026-07-18T21:10:36.409+00:00",
  "grade": 8.5,
  "agent_utility": 8.4,
  "price_usdc": 0.28,
  "stats": {
    "claims": 14,
    "entities": 22,
    "faq": 8,
    "downloads": 0,
    "paid_downloads": 0
  },
  "preview": {
    "faq_questions": [
      "What is the Industrial Dexterity Benchmark?",
      "What task does the paper evaluate?",
      "What success rate does the best policy achieve?",
      "How much better is the best configuration than the single-camera baseline?",
      "How many demonstrations are required?",
      "What is DAG-ROS?",
      "What is AG-iDP3?",
      "Does the paper prove learned policies beat classical robotics?"
    ],
    "entity_names": [
      {
        "name": "Industrial Dexterity Benchmark",
        "type": "benchmark"
      },
      {
        "name": "IDB",
        "type": "benchmark"
      },
      {
        "name": "IDB boards",
        "type": "hardware benchmark"
      },
      {
        "name": "Datacenter cable manipulation board",
        "type": "benchmark board"
      },
      {
        "name": "Datacenter cable management",
        "type": "task domain"
      },
      {
        "name": "Automotive cable harnesses",
        "type": "task domain"
      },
      {
        "name": "Gearbox assembly",
        "type": "task domain"
      },
      {
        "name": "DAG-ROS",
        "type": "framework"
      },
      {
        "name": "AG-iDP3",
        "type": "framework"
      },
      {
        "name": "Diffusion Policy",
        "type": "policy"
      },
      {
        "name": "Multimodal expansion Diffusion Policy",
        "type": "policy configuration"
      },
      {
        "name": "Single-camera RGB Diffusion Policy baseline",
        "type": "baseline"
      },
      {
        "name": "R3M encoder",
        "type": "encoder"
      },
      {
        "name": "RGB images",
        "type": "sensor modality"
      },
      {
        "name": "Multi-view RGB image source",
        "type": "sensor input"
      },
      {
        "name": "Point clouds",
        "type": "sensor modality"
      },
      {
        "name": "Joint positions",
        "type": "sensor modality"
      },
      {
        "name": "Wrist-frame wrench data",
        "type": "sensor modality"
      },
      {
        "name": "Teleoperated demonstrations",
        "type": "training data"
      },
      {
        "name": "Grasp and insert combined task success rate",
        "type": "metric"
      },
      {
        "name": "Classical modular robotics pipelines",
        "type": "method family"
      },
      {
        "name": "Classical vision and control robotic methods",
        "type": "method family"
      }
    ],
    "related_work_titles": [
      "Diffusion Policy variants for robot imitation learning",
      "R3M-style visual encoders for robot manipulation",
      "Classical modular robotics pipelines",
      "Teleoperation-based imitation learning"
    ],
    "application_industries": [
      "datacenter operations",
      "automotive manufacturing",
      "industrial assembly",
      "robotics R&D",
      "factory automation"
    ],
    "glossary_terms": [
      "Industrial Dexterity Benchmark",
      "Dexterous manipulation",
      "Classical modular robotics pipeline",
      "End-to-end policy",
      "Imitation learning",
      "Multimodal",
      "Diffusion Policy",
      "R3M encoder",
      "RGB images",
      "Point clouds",
      "Joint positions",
      "Wrist-frame wrench data",
      "Teleoperated demonstrations",
      "Datacenter cable manipulation board",
      "Grasp and insert combined task success rate"
    ]
  },
  "purchase": {
    "protocol": "x402",
    "version": 2,
    "network": "eip155:8453",
    "asset": "USDC",
    "price": 0.28,
    "full_markdown_url": "https://x402.disruptive-concepts.com/api/public/article/idb-multimodal-robot-cable-manipulation-benchmark-714021/full",
    "full_json_url": "https://x402.disruptive-concepts.com/api/public/article/idb-multimodal-robot-cable-manipulation-benchmark-714021/full?format=json"
  }
}