{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-10-01-006",
    "source_story_id": "tmp-story-stars-social-navigation-vlm-benchmark",
    "edition_id": "mp-2026-10-01-morning-0084",
    "edition_url": "https://themachinepress.com/edition/2026-10-01",
    "position": 6,
    "story_type": "dispatch",
    "section": "benchmarks-evals",
    "editorial_classification": "editorial",
    "headline": "Vision Models Lost to Rules in Social Navigation Scenes",
    "slug": "vision-models-lost-to-rules-in-social-navigation-scenes",
    "dek": "SocialNav-SUB tested spatial, temporal and social reasoning against human and rule-based baselines.",
    "summary": "SocialNav-SUB tested spatial, temporal and social reasoning against human and rule-based baselines.",
    "body_text": "The SocialNav-SUB benchmark asks vision-language models questions about real social-navigation scenes, including agent relationships and likely human intentions. Across the tested systems, the best model's agreement with human answers remained below both human consensus and a simpler rule-based approach. The result identifies a gap between general visual-language competence and the scene understanding needed for socially compliant robot motion. It evaluates perception and reasoning, not complete autonomous navigation safety.",
    "why_it_matters": "SocialNav-SUB tested spatial, temporal and social reasoning against human and rule-based baselines.",
    "limitations": [
      "It evaluates perception and reasoning, not complete autonomous navigation safety."
    ],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-01-006/vision-models-lost-to-rules-in-social-navigation-scenes",
    "json_url": "https://themachinepress.com/story/mp-2026-10-01-006.json",
    "first_published_at": "2026-10-01T09:00:00.000-04:00",
    "modified_at": "2026-10-01T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-10-01-006-001",
        "text": "SocialNav-SUB tested spatial, temporal and social reasoning against human and rule-based baselines.",
        "source_ids": [
          "source-2026-10-01-006"
        ],
        "qualification": "It evaluates perception and reasoning, not complete autonomous navigation safety."
      }
    ],
    "source_ids": [
      "source-2026-10-01-006"
    ],
    "tags": [
      "social navigation",
      "vision-language models",
      "benchmarking"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-10-01-006",
      "title": "arXiv preprint 2609.40245",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2609.40245",
      "canonical_url": "https://arxiv.org/abs/2609.40245",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-09-30T13:35:08.000-04:00",
      "accessed_at": "2026-10-01T08:18:00.000-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-10-01-006-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "Vision Models Lost to Rules in Social Navigation Scenes",
    "publisher": "The Machine Press",
    "published_at": "2026-10-01T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-01-006/vision-models-lost-to-rules-in-social-navigation-scenes"
  }
}