{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-11-06-005",
    "source_story_id": "tmp-story-oopsierecover",
    "edition_id": "mp-2026-11-06-morning-0119",
    "edition_url": "https://themachinepress.com/edition/2026-11-06",
    "position": 5,
    "story_type": "dispatch",
    "section": "robotics",
    "editorial_classification": "editorial",
    "headline": "A Few Corrections Teach the Reason for Failure",
    "slug": "a-few-corrections-teach-the-reason-for-failure",
    "dek": "OopsieRecover infers semantic intent from corrective demonstrations, then steers robot policies with spatial costs across new scenes.",
    "summary": "OopsieRecover infers semantic intent from corrective demonstrations, then steers robot policies with spatial costs across new scenes.",
    "body_text": "Instead of binding a correction to the exact observation in which it appeared, the framework uses abduction to infer why the original behavior was undesirable. It encodes that knowledge as dense spatial representations and conditional costs. Across six manipulation tasks and variations in objects, poses, scenes and related tasks, the authors report that a handful of demonstrations can transfer corrections without fresh data for every variation. The result remains confined to the paper’s tasks and vision-language-action models.",
    "why_it_matters": "OopsieRecover infers semantic intent from corrective demonstrations, then steers robot policies with spatial costs across new scenes.",
    "limitations": [],
    "importance": 9,
    "canonical_url": "https://themachinepress.com/story/mp-2026-11-06-005/a-few-corrections-teach-the-reason-for-failure",
    "json_url": "https://themachinepress.com/story/mp-2026-11-06-005.json",
    "first_published_at": "2026-11-06T09:00:00.000-05:00",
    "modified_at": "2026-11-06T09:00:00.000-05:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-11-06-005-001",
        "text": "OopsieRecover infers semantic intent from corrective demonstrations, then steers robot policies with spatial costs across new scenes.",
        "source_ids": [
          "source-2026-11-06-005"
        ],
        "qualification": null
      }
    ],
    "source_ids": [
      "source-2026-11-06-005"
    ],
    "tags": [
      "robot recovery",
      "vision-language-action",
      "corrective demonstrations"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-11-06-005",
      "title": "arXiv preprint 2611.08462, original v1",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2611.08462v1",
      "canonical_url": "https://arxiv.org/abs/2611.08462v1",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-11-05T13:53:16.000-05:00",
      "accessed_at": "2026-11-06T08:21:03.209-05:00",
      "supports_claim_ids": [
        "claim-mp-2026-11-06-005-001"
      ]
    }
  ],
  "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 Few Corrections Teach the Reason for Failure",
    "publisher": "The Machine Press",
    "published_at": "2026-11-06T09:00:00.000-05:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-11-06-005/a-few-corrections-teach-the-reason-for-failure"
  }
}