{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.31672", "kind": "paper", "url": "https://arxiv.org/abs/2606.31672"}], "authors": ["Ting-Bing Xu", "Jiacheng Sui", "Zhe Gao", "Kewei Shi", "Wenjin Yang", "Zhicheng Liu", "Zhaoxu Sun", "Mingchao Sun", "Hongyu Pan", "Fan Jiang", "Mu Xu", "Qi Fan", "Yang Gao", "Yong Li", "Baoquan Chen"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.31672", "source_url": "https://arxiv.org/abs/2606.31672"}, "categories": ["General AI", "Geometric reasoning", "Vision & 3D", "cs.CV"], "description": {"en": "WorldRoamBench evaluates interactive world models on long-horizon stability across action, vision, physics, and memory dimensions. It includes 600+ test cases in various scenes and views with continuous interaction."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.31672", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.31672", "title": "WorldRoamBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2606.31672", "modality": null, "name": "WorldRoamBench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "High", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "unknown", "origin_record_id": "bm_worldroambench_ba7d4093", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "9ceccd639d10b86b3417e6b66bf3cd613c249574dfdac36ba967b86a99c8e4ce", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.31672"}, "publisher": null, "released": "2026-06-30", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.31672", "source_url": "https://arxiv.org/abs/2606.31672"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-31672", "source": "claire_radar", "source_benchmark_id": "2606.31672", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Vision & 3D", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 0, "hfPaperUrl": "https://huggingface.co/papers/2606.31672", "observedAt": "2026-10-04T03:02:42.754940Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-10-04", "state": "stale"}}, "status": "stale"}, "capabilities": ["Geometric reasoning"], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "0151c8c8bd0d6bf6a0a8a303bcdef18953be429dbe8bd12944a57e965bb33af5", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "WorldRoamBench evaluates interactive world models on long-horizon stability across action, vision, physics, and memory dimensions. It includes 600+ test cases in various scenes and views with continuous interaction.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "We introduce WorldRoamBench, an open-world benchmark for long-horizon stability across four dimensions, each with tailored innovations: (i) Action: per-frame action metric bypassing cross-model semantic scale disparity and exposing failures hidden by trajectory; (ii) Vision: segment-based drift metric capturing non-monotonic mid-sequence collapse missed by start-vs-end comparisons; (iii) Physics: controllability-gated evaluation over mechanics, optics, and 3D consistency, scoring plausibility un"}, "familyId": "bmf_0d99b2d6873b", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_worldroambench_ba7d4093", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.31672", "paper": "https://arxiv.org/abs/2606.31672", "pdf": "https://arxiv.org/pdf/2606.31672", "project": null, "report": "https://arxiv.org/abs/2606.31672"}, "metrics": [], "motivation": "Despite rapid progress in interactive world models (IWMs), existing benchmarks evaluate action following only at trajectory level and ignore memory and interaction physics.", "name": "WorldRoamBench", "oneLine": "WorldRoamBench evaluates interactive world models on long-horizon stability across action, vision, physics, and memory dimensions. It includes 600+ test cases in various scenes and views with continuous interaction.", "paperTitle": "WorldRoamBench: An Open-World Benchmark for Long-Horizon Stability of Interactive World Models", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.31672", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-30", "source": {"authors": ["Ting-Bing Xu", "Jiacheng Sui", "Zhe Gao", "Kewei Shi", "Wenjin Yang", "Zhicheng Liu", "Zhaoxu Sun", "Mingchao Sun", "Hongyu Pan", "Fan Jiang", "Mu Xu", "Qi Fan", "Yang Gao", "Yong Li", "Baoquan Chen"], "categories": ["cs.CV", "cs.AI"], "id": "2606.31672", "title": "WorldRoamBench: An Open-World Benchmark for Long-Horizon Stability of Interactive World Models", "type": "arxiv", "url": "https://arxiv.org/abs/2606.31672"}, "sourceUpdatedAt": "2026-07-06T17:27:19Z", "topics": ["cs.CV"], "venueAttempts": [], "whyItMatters": "Interactive world models need stable, physically grounded, and memory-faithful behavior, but existing benchmarks ignore these aspects. WorldRoamBench provides a comprehensive evaluation revealing that no current model reliably satisfies all dimensions."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
