{"paper":{"title":"Perception with Guarantees: Certified Pose Estimation via Reachability Analysis","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Formal reachability analysis certifies bounds on 3D poses estimated from camera images of known targets.","cross_cats":["cs.RO"],"primary_cat":"cs.CV","authors_text":"Matthias Althoff, Tobias Ladner, Yasser Shoukry","submitted_at":"2026-02-10T17:55:49Z","abstract_excerpt":"Agents in cyber-physical systems are increasingly entrusted with safety-critical tasks. Ensuring safety of these agents often requires localizing the pose for subsequent actions. Pose estimates can, e.g., be obtained from various combinations of lidar sensors, cameras, and external services such as GPS. Crucially, in safety-critical domains, a rough estimate is insufficient to formally determine safety, i.e., guaranteeing safety even in the worst-case scenario, and external services might additionally not be trustworthy. We address this problem by presenting a certified pose estimation in 3D s"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"presenting a certified pose estimation in 3D solely from a camera image and a well-known target geometry. This is realized by formally bounding the pose, which is computed by leveraging recent results from reachability analysis and formal neural network verification.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The target geometry is perfectly known and modeled, and the formal verification of the neural network produces sufficiently tight bounds that remain practical for real-time safety-critical decisions.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Certified 3D pose estimation from camera images using reachability analysis and formal NN verification delivers formal bounds for safety-critical localization.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Formal reachability analysis certifies bounds on 3D poses estimated from camera images of known targets.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"19c2f9309c3c58563cf73e51fc53c848398a2bbc543192a069c450a5f7b7cf4c"},"source":{"id":"2602.10032","kind":"arxiv","version":2},"verdict":{"id":"a6156a34-7557-462b-b940-f616bd73228f","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T02:33:52.946368Z","strongest_claim":"presenting a certified pose estimation in 3D solely from a camera image and a well-known target geometry. This is realized by formally bounding the pose, which is computed by leveraging recent results from reachability analysis and formal neural network verification.","one_line_summary":"Certified 3D pose estimation from camera images using reachability analysis and formal NN verification delivers formal bounds for safety-critical localization.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The target geometry is perfectly known and modeled, and the formal verification of the neural network produces sufficiently tight bounds that remain practical for real-time safety-critical decisions.","pith_extraction_headline":"Formal reachability analysis certifies bounds on 3D poses estimated from camera images of known targets."},"references":{"count":51,"sample":[{"doi":"","year":2023,"title":"Journal of the Franklin Institute (2023)","work_id":"a930933b-3e32-454c-970c-4b1621a5a4ab","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2010,"title":"Althoff, M.: Reachability analysis and its application to the safety assessment of autonomous cars. Ph.D. thesis, Technische Universität München (2010)","work_id":"1fe497df-6df6-4aa0-bc9c-6240b3ba129e","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2015,"title":"Althoff, M.: An introduction to CORA 2015. In: Proc. of the 1st and 2nd Workshop on Applied Verification for Continuous and Hybrid Systems. pp. 120–151 (2015)","work_id":"6c90a6f5-7b93-4bbf-a330-413dbd57e52e","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Foundations and Trends in Machine Learning (2023)","work_id":"e9db70b7-4c44-4621-b56f-81c3b70dd793","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Boldmethod: Runway stripes and markings, explained. (2025), https://www.boldmethod.com/learn-to-fly/regulations/ runway-markings-and-spacing/","work_id":"9e569bef-0cae-4172-83ae-6d802f3e99ba","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":51,"snapshot_sha256":"be3c807e3c9d1604969ef9981cba32c530b7d9d0a0664f147beb9836132eba87","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"1994a354db007e2846add450b6b9501a44b9d5fb958d4da27c9132f49eb39dfb"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}