{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2605.23176", "kind": "paper", "url": "https://arxiv.org/abs/2605.23176"}], "authors": ["Hao Vo", "Khoa Vo", "Phu Loc Nguyen", "Sieu Tran", "Duc Minh Nguyen", "Ngo Xuan Cuong", "Gladys Gawugah", "Sreevenkata Anjani Tishita Godavarthi", "Chase Rainwater", "Nghi D. Q. Bui", "Anh Nguyen", "Duy Minh Ho Nguyen", "Ngan Le"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2605.23176", "source_url": "https://arxiv.org/abs/2605.23176"}, "categories": ["Autonomous Driving", "Vision & 3D", "cs.CV"], "description": {"en": "DriveSpatial is a benchmark of 15.6K human-verified QA pairs across 20 tasks from five AD datasets, evaluating VLMs on cognitive scene construction, multi-view relational understanding, temporal reasoning, and generalization."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2605.23176", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2605.23176", "title": "DRIVESPATIAL"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2605.23176", "modality": null, "name": "DRIVESPATIAL", "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_drivespatial_f218db55", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "f981de3f7a946e302973debd2d2828b82511c50fd52426e1311856c8865f9dec", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2605.23176"}, "publisher": null, "released": "2026-05-22", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2605.23176", "source_url": "https://arxiv.org/abs/2605.23176"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2605-23176", "source": "claire_radar", "source_benchmark_id": "2605.23176", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["Autonomous Driving"], "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": null, "hfPaperUrl": "https://huggingface.co/papers/2605.23176", "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", "state": "unavailable"}}, "status": "unavailable"}, "capabilities": [], "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": "0a5d90c49b179b9c945fd98805ee2f9c3e4a362e7f16159a1657981f60a0f746", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "DriveSpatial is a benchmark of 15.6K human-verified QA pairs across 20 tasks from five AD datasets, evaluating VLMs on cognitive scene construction, multi-view relational understanding, temporal reasoning, and generalization.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "We introduce DriveSpatial, a benchmark of 15.6K human-verified QA pairs across 20 tasks from five large-scale AD datasets."}, "familyId": "bmf_f22a23f585f9", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_drivespatial_f218db55", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": ["Automotive"], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2605.23176", "paper": "https://arxiv.org/abs/2605.23176", "pdf": "https://arxiv.org/pdf/2605.23176", "project": null, "report": "https://arxiv.org/abs/2605.23176"}, "metrics": [], "motivation": "Spatiotemporal intelligence in autonomous driving (AD) requires an agent to integrate multi-view observations into a coherent scene representation, maintain object continuity across viewpoints and time, and reason about spatial relations, interactions, and future dynamics.", "name": "DRIVESPATIAL", "oneLine": "DriveSpatial is a benchmark of 15.6K human-verified QA pairs across 20 tasks from five AD datasets, evaluating VLMs on cognitive scene construction, multi-view relational understanding, temporal reasoning, and generalization.", "paperTitle": "DRIVESPATIAL: A Benchmark for Spatiotemporal Intelligence in VLMs for Autonomous Driving", "primaryDomain": "Autonomous Driving", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2605.23176", "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": "2026-05-22", "source": {"authors": ["Hao Vo", "Khoa Vo", "Phu Loc Nguyen", "Sieu Tran", "Duc Minh Nguyen", "Ngo Xuan Cuong", "Gladys Gawugah", "Sreevenkata Anjani Tishita Godavarthi", "Chase Rainwater", "Nghi D. Q. Bui", "Anh Nguyen", "Duy Minh Ho Nguyen", "Ngan Le"], "categories": ["cs.CV"], "id": "2605.23176", "title": "DRIVESPATIAL: A Benchmark for Spatiotemporal Intelligence in VLMs for Autonomous Driving", "type": "arxiv", "url": "https://arxiv.org/abs/2605.23176"}, "sourceUpdatedAt": "2026-06-15T18:43:33Z", "topics": ["cs.CV"], "venueAttempts": [], "whyItMatters": "Existing AD vision-language benchmarks focus on static, single-view QA, leaving unclear whether VLMs can reason over dynamic driving scenes. DriveSpatial provides a multi-sourced, human-verified protocol to measure spatiotemporal reasoning and reveals a substantial human-model gap."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
