{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-09-24-008",
    "source_story_id": "tmp-story-pointcast-unified-object-world-model",
    "edition_id": "mp-2026-09-24-morning-0077",
    "edition_url": "https://themachinepress.com/edition/2026-09-24",
    "position": 8,
    "story_type": "dispatch",
    "section": "robotics",
    "editorial_classification": "editorial",
    "headline": "One Robot World Model Handled Cabinets, Rope and Cloth",
    "slug": "one-robot-world-model-handled-cabinets-rope-and-cloth",
    "dek": "PointCast tracks identified 3D points instead of committing to one object's mesh or topology.",
    "summary": "PointCast tracks identified 3D points instead of committing to one object's mesh or topology.",
    "body_text": "PointCast predicts future trajectories for sets of identified 3D points on objects and a robot end effector. Separate checkpoints using the same 19.8-million-parameter architecture were best on three of four simulated regimes and second on rigid objects; on a real teleoperation dataset, it had the lowest mean error in four of six categories. The work unifies a representation and training recipe, not one universal checkpoint for every object.",
    "why_it_matters": "PointCast tracks identified 3D points instead of committing to one object's mesh or topology.",
    "limitations": [
      "The work unifies a representation and training recipe, not one universal checkpoint for every object."
    ],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-24-008/one-robot-world-model-handled-cabinets-rope-and-cloth",
    "json_url": "https://themachinepress.com/story/mp-2026-09-24-008.json",
    "first_published_at": "2026-09-24T09:00:00.000-04:00",
    "modified_at": "2026-09-24T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-09-24-008-001",
        "text": "PointCast tracks identified 3D points instead of committing to one object's mesh or topology.",
        "source_ids": [
          "source-2026-09-24-008"
        ],
        "qualification": "The work unifies a representation and training recipe, not one universal checkpoint for every object."
      }
    ],
    "source_ids": [
      "source-2026-09-24-008"
    ],
    "tags": [
      "world models",
      "robot manipulation",
      "3D prediction"
    ],
    "image_url": "https://themachinepress.com/issues/2026-09-24/pointcast-robot-arm-file.webp",
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-09-24-008",
      "title": "arXiv preprint 2609.28393",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2609.28393",
      "canonical_url": "https://arxiv.org/abs/2609.28393",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-09-23T13:02:18.000-04:00",
      "accessed_at": "2026-09-24T08:17:17.000-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-09-24-008-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "One Robot World Model Handled Cabinets, Rope and Cloth",
    "publisher": "The Machine Press",
    "published_at": "2026-09-24T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-24-008/one-robot-world-model-handled-cabinets-rope-and-cloth"
  }
}