{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.27079", "kind": "paper", "url": "https://arxiv.org/abs/2606.27079"}], "authors": ["Mingyang Lyu", "Yinqian Sun", "Yiyang Jia", "Sicheng Shen", "Moquan Sha", "Huangrui Li", "Feifei Zhao", "Yi Zeng"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.27079", "source_url": "https://arxiv.org/abs/2606.27079"}, "categories": ["Multimodal", "Robotics & Embodied AI", "Safety", "Safety & Trustworthiness"], "description": {"en": "A diagnostic safety benchmark for vision-language-action models, evaluating policies across physical interaction, instruction, and perception safety with cumulative cost and risk exposure metrics."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.27079", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.27079", "title": "ForesightSafety-VLA"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Institute of Automation", "University of Chinese Academy of Sciences"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2606.27079", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2606.27079", "source_url": "https://openalex.org/W7166132482"}, "key": "claire-radar:2606.27079", "modality": null, "name": "ForesightSafety-VLA", "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_foresightsafety-vla_a9025fe7", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "6ade5c3eaf34825e93ca6fa9a7000f5a84737c1003e62560694439d5418d6ef4", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.27079"}, "publisher": null, "released": "2026-06-25", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.27079", "source_url": "https://arxiv.org/abs/2606.27079"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-27079", "source": "claire_radar", "source_benchmark_id": "2606.27079", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Mixed", "applicationDomains": ["Robotics & Embodied AI"], "area": "Safety & Trustworthiness", "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/2606.27079", "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": "Aggregate Existing", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "91363d178a06d8cad577c721e4b94b919b1f532e69b5384ec53f38bacda3980f", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "A diagnostic safety benchmark for vision-language-action models, evaluating policies across physical interaction, instruction, and perception safety with cumulative cost and risk exposure metrics.", "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": "To address this gap, we introduce ForesightSafety-VLA, a diagnostic benchmark that makes safety the primary evaluation target for VLA systems."}, "familyId": "bmf_f09eeaf1bfcf", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_foresightsafety-vla_a9025fe7", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": ["Robotics"], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.27079", "paper": "https://arxiv.org/abs/2606.27079", "pdf": "https://arxiv.org/pdf/2606.27079", "project": null, "report": "https://arxiv.org/abs/2606.27079"}, "metrics": [], "motivation": "In embodied intelligence, safety is a prerequisite for reliable robot deployment in the physical world.", "name": "ForesightSafety-VLA", "oneLine": "A diagnostic safety benchmark for vision-language-action models, evaluating policies across physical interaction, instruction, and perception safety with cumulative cost and risk exposure metrics.", "paperTitle": "ForesightSafety-VLA: A Unified Diagnostic Safety Benchmark for Vision-Language-Action Models", "primaryDomain": "Robotics & Embodied AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.27079", "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": "2026-06-25", "source": {"authors": ["Mingyang Lyu", "Yinqian Sun", "Yiyang Jia", "Sicheng Shen", "Moquan Sha", "Huangrui Li", "Feifei Zhao", "Yi Zeng"], "categories": ["cs.RO"], "id": "2606.27079", "title": "ForesightSafety-VLA: A Unified Diagnostic Safety Benchmark for Vision-Language-Action Models", "type": "arxiv", "url": "https://arxiv.org/abs/2606.27079"}, "sourceUpdatedAt": "2026-06-27T13:41:35Z", "topics": ["Multimodal", "Safety"], "venueAttempts": [], "whyItMatters": "The benchmark addresses the lack of systematic safety evaluation for embodied VLA models, offering process-level risk metrics and a taxonomy to localize failure sources, which supports model selection and safety improvements."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
