{"paper":{"title":"NuPlan: A closed-loop ML-based planning benchmark for autonomous vehicles","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"NuPlan establishes the first closed-loop benchmark for machine learning planners in autonomous driving.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alex Lang, Eric Wolff, Holger Caesar, Juraj Kabzan, Kok Seang Tan, Luke Fletcher, Oscar Beijbom, Sammy Omari, Whye Kit Fong","submitted_at":"2021-06-22T14:24:55Z","abstract_excerpt":"In this work, we propose the world's first closed-loop ML-based planning benchmark for autonomous driving. While there is a growing body of ML-based motion planners, the lack of established datasets and metrics has limited the progress in this area. Existing benchmarks for autonomous vehicle motion prediction have focused on short-term motion forecasting, rather than long-term planning. This has led previous works to use open-loop evaluation with L2-based metrics, which are not suitable for fairly evaluating long-term planning. Our benchmark overcomes these limitations by introducing a large-s"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"In this work, we propose the world's first closed-loop ML-based planning benchmark for autonomous driving.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the chosen metrics and reactive-agent simulator will produce rankings that correlate with real-world safety and performance once deployed on physical vehicles.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"NuPlan is the first closed-loop benchmark for ML-based autonomous vehicle planning, with 1500h multi-city driving data, reactive simulation, and scenario-specific metrics.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"NuPlan establishes the first closed-loop benchmark for machine learning planners in autonomous driving.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"975e888e3c7661148de1dda04310d4319b4febc572c6fa8b77eff698bdf1096e"},"source":{"id":"2106.11810","kind":"arxiv","version":4},"verdict":{"id":"93df7ad5-d461-4b9d-af8d-a84d91ccee7a","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T08:57:39.837798Z","strongest_claim":"In this work, we propose the world's first closed-loop ML-based planning benchmark for autonomous driving.","one_line_summary":"NuPlan is the first closed-loop benchmark for ML-based autonomous vehicle planning, with 1500h multi-city driving data, reactive simulation, and scenario-specific metrics.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the chosen metrics and reactive-agent simulator will produce rankings that correlate with real-world safety and performance once deployed on physical vehicles.","pith_extraction_headline":"NuPlan establishes the first closed-loop benchmark for machine learning planners in autonomous driving."},"references":{"count":24,"sample":[{"doi":"","year":2017,"title":"CommonRoad: Composable benchmarks for motion plan- ning on roads","work_id":"25af57e2-ac60-41a9-bded-f3495a3658f9","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"Chauf- feurnet: Learning to drive by imitating the best and synthe- sizing the worst","work_id":"de43fcae-ee6a-49c5-b229-01a0ce8e9804","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Learning to drive from simulation without real world labels","work_id":"6a3833a4-2967-48ba-b3df-f714fddced35","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"Lang, Sourabh V ora, Venice Erin Liong, Qiang Xu, Anush Krishnan, Yu Pan, Gi- ancarlo Baldan, and Oscar Beijbom","work_id":"de934989-2ff8-49f5-a3b3-0fe39f4a14c2","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"MP3: A unified model to map, perceive, predict and plan","work_id":"ffea778c-e8d9-4a0e-b307-aed0f162ae7d","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":24,"snapshot_sha256":"c44d7de0fc6611a2c42f361a188d449d6663762170b4e5f8b94b9bb1332021e3","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"415d64bdf6099cc2c8f1e608edf4435de561ce19819a59cd2a2953729d095917"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}