{"record": {"aliases": ["Rule-Compliant Visual Spatial Planning for Multimodal Large Language Models"], "artifacts": [{"id": "arxiv:2608.20237", "kind": "paper", "url": "https://arxiv.org/abs/2608.20237"}, {"id": "gh:oceanflowlab/rulemaze", "kind": "repo", "url": "https://github.com/oceanflowlab/RuleMaze"}], "authors": ["Yu Chen", "Ting Lei", "Yaoyi Li", "Jia Cai", "Zhecen Wu", "Yang Liu"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.20237", "source_url": "https://arxiv.org/abs/2608.20237"}, "categories": ["General AI", "Multimodal", "Planning"], "description": {"en": "Benchmark for rule-compliant visual spatial planning in multimodal LLMs, requiring maze navigation under natural-language rules with automated rule generation and validation."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.20237", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.20237", "title": "RuleMaze"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.20237", "modality": null, "name": "RuleMaze", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "true", "origin_record_id": "bm_rule-compliant-visual-spatial-planning-for_2793101a", "origin_source": "arxiv", "recognition_confidence": "0.55", "record_sha256": "4bd3468ed4eae3e4f43a9010a91d70a7ed1a631a49582b09b14ecdd3106a24bc", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-08-25T17:43:56.718736Z", "source_url": "https://arxiv.org/abs/2608.20237"}, "publisher": null, "released": "2026-08-20", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.20237", "source_url": "https://arxiv.org/abs/2608.20237"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-20237", "source": "claire_radar", "source_benchmark_id": "2608.20237", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": ["Rule-Compliant Visual Spatial Planning for Multimodal Large Language Models"], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Multimodal", "attention": {"asOf": "2026-08-26", "attemptedAt": "2026-08-25T16:48:55.714492Z", "githubRepo": "https://github.com/oceanflowlab/RuleMaze", "githubScope": "benchmark_repo", "githubStars": 2, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2608.20237", "observedAt": "2026-08-25T16:48:55.714492Z", "signalStatus": {"githubStars": {"observedAt": "2026-08-25T16:48:55.714492Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-08-25T16:48:55.714492Z", "state": "unavailable"}}, "status": "fresh"}, "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "Multimodal LLM planning is a hot topic and the benchmark provides code, dataset, and a project page, likely generating broad interest.", "score": 70}, "capabilities": ["Planning"], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-25T17:43:56.718736Z", "inputHash": "9e4db56f22d9ace1af1ad77b037cf7dd1825cd943b50bc4b9777ab926d772925", "model": "deepseek-v4-pro"}, "curation": {"canonicalNameEvidence": "we introduce RuleMaze, a controllable benchmark in which MLLMs must navigate mazes while obeying natural-language rules", "canonicalNameSource": "abstract", "decisionReason": "The paper introduces a benchmark with code, dataset, and evaluation pipeline, supporting ongoing model comparison.", "model": "deepseek-v4-pro", "reviewedAt": "2026-08-25T17:43:56.718736Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Benchmark for rule-compliant visual spatial planning in multimodal LLMs, requiring maze navigation under natural-language rules with automated rule generation and validation.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "score_submission", "evidence": {"reasonCodes": ["benchmark term in abstract", "benchmark release stated in one sentence", "public artifact URL"], "snippet": "To address this gap, we introduce RuleMaze, a controllable benchmark in which MLLMs must navigate mazes while obeying natural-language rules of varying complexity."}, "familyId": "bmf_1b6b5724735c", "firstSeenAt": "2026-08-21", "heat": null, "id": "bm_rule-compliant-visual-spatial-planning-for_2793101a", "indexedAt": "2026-08-21T04:30:09.569171Z", "industrySectors": [], "links": {"code": "https://github.com/oceanflowlab/RuleMaze", "data": null, "paper": "https://arxiv.org/abs/2608.20237", "pdf": "https://arxiv.org/pdf/2608.20237", "project": null, "report": "https://arxiv.org/abs/2608.20237"}, "metrics": [], "motivation": "Multimodal large language models (MLLMs) combine linguistic reasoning with visual perception, yet their ability to perform visual spatial planning under explicit or previously unseen rule constraints remains underexplored.", "name": "RuleMaze", "oneLine": "Benchmark for rule-compliant visual spatial planning in multimodal LLMs, requiring maze navigation under natural-language rules with automated rule generation and validation.", "paperTitle": "Rule-Compliant Visual Spatial Planning for Multimodal Large Language Models", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.20237", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-21T04:30:09.569171Z"}, "publications": [], "publishers": [{"name": "OceanFlowLab", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://github.com/oceanflowlab/RuleMaze"}], "ranking": {"30d": {"confidence": "Low", "coverage": 0.55, "growth": {"components": {"githubStars": {"percentile": null, "value": null}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 30}, "level": {"components": {"githubStars": {"percentile": 0.5714285714285714, "value": 2}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 53, "score": 54}, "method": "attention-ranking-v11", "rank": 53, "score": 54}, "90d": {"confidence": "Low", "coverage": 0.55, "growth": {"components": {"githubStars": {"percentile": null, "value": null}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"percentile": 0.3802228412256267, "value": 2}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 236, "score": 43}, "method": "attention-ranking-v11", "rank": 236, "score": 43}}, "readiness": "Runnable", "recognitionConfidence": 0.55, "relation": "introduces", "releasedAt": "2026-08-20", "source": {"authors": ["Yu Chen", "Ting Lei", "Yaoyi Li", "Jia Cai", "Zhecen Wu and Yang Liu"], "categories": ["cs.AI"], "id": "2608.20237", "title": "Rule-Compliant Visual Spatial Planning for Multimodal Large Language Models", "type": "arxiv", "url": "https://arxiv.org/abs/2608.20237"}, "sourceUpdatedAt": "2026-08-20T16:28:28Z", "topics": ["Multimodal"], "venueAttempts": [], "whyItMatters": "Fills a gap in evaluating MLLMs on joint visual perception, rule interpretation, and constrained action planning, with scalable rule construction and a public leaderboard."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
