{"paper":{"title":"Enhancing End-to-End Autonomous Driving with Latent World Model","license":"http://creativecommons.org/licenses/by/4.0/","headline":"LAW uses self-supervised prediction of future scene features to strengthen end-to-end autonomous driving planners.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Jiawei He, Lue Fan, Tieniu Tan, Yingyan Li, Yuntao Chen, Yuqi Wang, Zhaoxiang Zhang","submitted_at":"2024-06-12T17:59:21Z","abstract_excerpt":"In autonomous driving, end-to-end planners directly utilize raw sensor data, enabling them to extract richer scene features and reduce information loss compared to traditional planners. This raises a crucial research question: how can we develop better scene feature representations to fully leverage sensor data in end-to-end driving? Self-supervised learning methods show great success in learning rich feature representations in NLP and computer vision. Inspired by this, we propose a novel self-supervised learning approach using the LAtent World model (LAW) for end-to-end driving. LAW predicts "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"LAW achieves state-of-the-art performance across multiple benchmarks, including real-world open-loop benchmark nuScenes, NAVSIM, and simulator-based closed-loop benchmark CARLA.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the self-supervised future-feature prediction task will reliably improve downstream trajectory prediction quality in both open-loop and closed-loop settings without introducing new failure modes.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"LAW introduces a self-supervised prediction task on latent scene features that boosts end-to-end driving performance on nuScenes, NAVSIM, and CARLA benchmarks.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"LAW uses self-supervised prediction of future scene features to strengthen end-to-end autonomous driving planners.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"2a6709dbc9289830d510a8fea4ced0ab7c1e3e66003058b8b8d0584f8b1356e8"},"source":{"id":"2406.08481","kind":"arxiv","version":2},"verdict":{"id":"6aad3d9b-fa24-4256-b8d5-cdd546e7e996","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-17T07:33:05.200378Z","strongest_claim":"LAW achieves state-of-the-art performance across multiple benchmarks, including real-world open-loop benchmark nuScenes, NAVSIM, and simulator-based closed-loop benchmark CARLA.","one_line_summary":"LAW introduces a self-supervised prediction task on latent scene features that boosts end-to-end driving performance on nuScenes, NAVSIM, and CARLA benchmarks.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the self-supervised future-feature prediction task will reliably improve downstream trajectory prediction quality in both open-loop and closed-loop settings without introducing new failure modes.","pith_extraction_headline":"LAW uses self-supervised prediction of future scene features to strengthen end-to-end autonomous driving planners."},"references":{"count":21,"sample":[{"doi":"","year":null,"title":"NuPlan: A closed-loop ML-based planning benchmark for autonomous vehicles","work_id":"b16aace6-deff-4546-b333-bcb7c9c07cdb","ref_index":1,"cited_arxiv_id":"2106.11810","is_internal_anchor":true},{"doi":"","year":null,"title":"Navsim: Data-driven non-reactive autonomous vehicle simulation and benchmarking","work_id":"9f4cc1ee-32ef-4029-9b25-a40ec78a3df3","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding","work_id":"ed240a10-5b19-406c-baa5-30803f465785","ref_index":3,"cited_arxiv_id":"1810.04805","is_internal_anchor":true},{"doi":"","year":2025,"title":"GAIA-1: A Generative World Model for Autonomous Driving","work_id":"313484e6-a442-4522-8e19-d07e502844a8","ref_index":4,"cited_arxiv_id":"2309.17080","is_internal_anchor":true},{"doi":"","year":null,"title":"Planning-oriented autonomous driving","work_id":"781f0902-b7d6-482e-b8f0-a4fb3ad09dfd","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":21,"snapshot_sha256":"6e98e44fdccf0448eb780c50b95d6ee9b95f21c7238b40599e87910f6fafe425","internal_anchors":7},"formal_canon":{"evidence_count":2,"snapshot_sha256":"38a2ee7493751d7d08af52ba67eecabf0499e4f392353e3563971fd58716c972"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}