{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-10-10-008",
    "source_story_id": "tmp-story-success-guided-robot-rl",
    "edition_id": "mp-2026-10-10-morning-0092",
    "edition_url": "https://themachinepress.com/edition/2026-10-10",
    "position": 8,
    "story_type": "dispatch",
    "section": "robotics",
    "editorial_classification": "editorial",
    "headline": "A Million Simulators Learned More by Ignoring Easy Resets",
    "slug": "a-million-simulators-learned-more-by-ignoring-easy-resets",
    "dek": "Success Guided Sampling concentrates robot training near the frontier between mastered and not-yet-reachable task configurations.",
    "summary": "Success Guided Sampling concentrates robot training near the frontier between mastered and not-yet-reachable task configurations.",
    "body_text": "Uniformly sampling simulator resets wastes training on situations a robot has already mastered or cannot yet attempt. Success Guided Sampling adapts the reset distribution toward configurations around the policy’s current capability frontier. In experiments scaling to more than one million parallel environments, that change let reinforcement learning solve multi-terrain locomotion and contact-rich assembly tasks that the tested prior methods did not solve. The authors then distilled manipulation policies into RGB-based controllers and report zero-shot transfer to real hardware. The paper’s claim is about using each simulated batch more effectively; it does not show that raw simulator count alone produces general-purpose robot control.",
    "why_it_matters": "Success Guided Sampling concentrates robot training near the frontier between mastered and not-yet-reachable task configurations.",
    "limitations": [
      "Uniformly sampling simulator resets wastes training on situations a robot has already mastered or cannot yet attempt.",
      "In experiments scaling to more than one million parallel environments, that change let reinforcement learning solve multi-terrain locomotion and contact-rich assembly tasks that the tested prior methods did not solve.",
      "The paper’s claim is about using each simulated batch more effectively; it does not show that raw simulator count alone produces general-purpose robot control."
    ],
    "importance": 9,
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-10-008/a-million-simulators-learned-more-by-ignoring-easy-resets",
    "json_url": "https://themachinepress.com/story/mp-2026-10-10-008.json",
    "first_published_at": "2026-10-10T09:00:00.000-04:00",
    "modified_at": "2026-10-10T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-10-10-008-001",
        "text": "Success Guided Sampling concentrates robot training near the frontier between mastered and not-yet-reachable task configurations.",
        "source_ids": [
          "source-2026-10-10-008"
        ],
        "qualification": "Uniformly sampling simulator resets wastes training on situations a robot has already mastered or cannot yet attempt."
      }
    ],
    "source_ids": [
      "source-2026-10-10-008"
    ],
    "tags": [
      "robot learning",
      "reinforcement learning",
      "simulation scaling"
    ],
    "image_url": "https://themachinepress.com/issues/2026-10-10/success-guided-compute.webp",
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-10-10-008",
      "title": "arXiv preprint 2610.12465, original v1",
      "publisher": "arXiv / submitting research authors",
      "url": "https://arxiv.org/abs/2610.12465v1",
      "canonical_url": "https://arxiv.org/abs/2610.12465v1",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-10-08T13:59:50.000-04:00",
      "accessed_at": "2026-10-10T08:22:13.309-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-10-10-008-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "A Million Simulators Learned More by Ignoring Easy Resets",
    "publisher": "The Machine Press",
    "published_at": "2026-10-10T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-10-008/a-million-simulators-learned-more-by-ignoring-easy-resets"
  }
}