{"paper":{"title":"Vision-Based Safe Human-Robot Collaboration with Uncertainty Guarantees","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Conformal prediction sets deliver valid high-confidence bounds on human motion predictions for integration into certifiable robot safety systems.","cross_cats":["cs.CV"],"primary_cat":"cs.RO","authors_text":"Jakob Thumm, Marco Pavone, Marian Frei, Matthias Althoff, Tianle Ni","submitted_at":"2026-04-16T16:53:26Z","abstract_excerpt":"We propose a framework for vision-based human pose estimation and motion prediction that gives conformal prediction guarantees for certifiably safe human-robot collaboration. Our framework combines aleatoric uncertainty estimation with OOD detection for high probabilistic confidence. To integrate our pipeline in certifiable safety frameworks, we propose conformal prediction sets for human motion predictions with high, valid confidence. We evaluate our pipeline on recorded human motion data and a real-world human-robot collaboration setting."},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We propose conformal prediction sets for human motion predictions with high, valid confidence that can be integrated into certifiable safety frameworks for human-robot collaboration.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that the conformal prediction sets remain valid when deployed in real-world human-robot settings, which requires the test distribution of human motions to be sufficiently similar to the calibration data used to construct the sets.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Proposes a vision-based human pose estimation and motion prediction pipeline that uses conformal prediction sets to provide valid confidence guarantees for safe human-robot collaboration.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Conformal prediction sets deliver valid high-confidence bounds on human motion predictions for integration into certifiable robot safety systems.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"faba6af5c7a6d43460b944b6dc314800330975ef21023413f93989ae767c26ea"},"source":{"id":"2604.15221","kind":"arxiv","version":2},"verdict":{"id":"bb0459d8-0672-4047-a7b6-d81ebef93fa3","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T17:14:58.104200Z","strongest_claim":"We propose conformal prediction sets for human motion predictions with high, valid confidence that can be integrated into certifiable safety frameworks for human-robot collaboration.","one_line_summary":"Proposes a vision-based human pose estimation and motion prediction pipeline that uses conformal prediction sets to provide valid confidence guarantees for safe human-robot collaboration.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that the conformal prediction sets remain valid when deployed in real-world human-robot settings, which requires the test distribution of human motions to be sufficiently similar to the calibration data used to construct the sets.","pith_extraction_headline":"Conformal prediction sets deliver valid high-confidence bounds on human motion predictions for integration into certifiable robot safety systems."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.15221/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":36,"sample":[{"doi":"","year":2012,"title":"On making robots understand safety: Embedding injury knowledge into control,","work_id":"75204881-f8b4-4964-8044-d5d688d04b74","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2017,"title":"Online verification of multiple safety criteria for a robot trajectory,","work_id":"9800b9c2-44d7-4086-b0cb-2827898704a7","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Provably safe deep reinforcement learning for robotic manipulation in human environments,","work_id":"7af6754c-19ed-470e-8f89-b56b97c33584","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2026,"title":"A general safety framework for autonomous manipulation in human environments,","work_id":"f35c297c-a80e-4f2b-80e3-66bae8ad766d","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2016,"title":"Safety in human-robot collaborative manufacturing environments: Metrics and control,","work_id":"b97c3933-3c8f-49c4-bf87-25f6c3762f41","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":36,"snapshot_sha256":"555e68a5e04c6d5e441d98ac847dd3474c7acdcd7037ba0dc276b8d856e8d30f","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"7f52609d538bd73e3828d3d0de6b9786b919dadfb07cd6f782447876612351d2"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}