{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-09-20-010",
    "source_story_id": "tmp-story-gavel-graph-world-model",
    "edition_id": "mp-2026-09-20-morning-0073",
    "edition_url": "https://themachinepress.com/edition/2026-09-20",
    "position": 10,
    "story_type": "dispatch",
    "section": "robotics",
    "editorial_classification": "editorial",
    "headline": "A Graph Checked the Robot’s Plan Before It Moved",
    "slug": "a-graph-checked-the-robot-s-plan-before-it-moved",
    "dek": "GAVEL lifted one compact model from 41.2% to 91.8% success on long-horizon simulated tasks.",
    "summary": "GAVEL lifted one compact model from 41.2% to 91.8% success on long-horizon simulated tasks.",
    "body_text": "GAVEL represents object relations, action preconditions, effects and beliefs about hidden object locations in an explicit graph world model. It predicts the consequences of language-model actions, repairs violations that follow from the graph and reserves language-model replanning for semantic errors. On BEHAVIOR-1K with Qwen3-8B, single-task success rose from 41.2% to 91.8% and multi-task success from 19.9% to 92.6%; belief-aware reordering cut travel about 5.4% versus a static variant. These are embodied-simulation results, not household-robot deployment evidence.",
    "why_it_matters": "GAVEL lifted one compact model from 41.2% to 91.8% success on long-horizon simulated tasks.",
    "limitations": [
      "These are embodied-simulation results, not household-robot deployment evidence."
    ],
    "importance": 9,
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-20-010/a-graph-checked-the-robot-s-plan-before-it-moved",
    "json_url": "https://themachinepress.com/story/mp-2026-09-20-010.json",
    "first_published_at": "2026-09-20T09:00:00.000-04:00",
    "modified_at": "2026-09-20T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-09-20-010-001",
        "text": "GAVEL lifted one compact model from 41.2% to 91.8% success on long-horizon simulated tasks.",
        "source_ids": [
          "source-2026-09-20-010"
        ],
        "qualification": "These are embodied-simulation results, not household-robot deployment evidence."
      }
    ],
    "source_ids": [
      "source-2026-09-20-010"
    ],
    "tags": [
      "robot planning",
      "world models",
      "verification"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-09-20-010",
      "title": "arXiv preprint 2609.19315",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2609.19315",
      "canonical_url": "https://arxiv.org/abs/2609.19315",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-09-16T14:25:58.000-04:00",
      "accessed_at": "2026-09-20T08:23:59.913-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-09-20-010-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "A Graph Checked the Robot’s Plan Before It Moved",
    "publisher": "The Machine Press",
    "published_at": "2026-09-20T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-20-010/a-graph-checked-the-robot-s-plan-before-it-moved"
  }
}