{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:47NZQJKUXNV7TJGINCL7Q6IAUZ","short_pith_number":"pith:47NZQJKU","schema_version":"1.0","canonical_sha256":"e7db982554bb6bf9a4c86897f87900a65f1df2efe5b814e61e62edc1a10bb6de","source":{"kind":"arxiv","id":"2508.06571","version":3},"attestation_state":"computed","paper":{"title":"IRL-VLA: Training an Vision-Language-Action Policy via Reward World Model","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","cs.RO"],"primary_cat":"cs.AI","authors_text":"Anqing Jiang, Hao Jiang, Hao Sun, Jijun Wang, Jinhao Chai, Lijuan Zhu, Li Sun, Shichen Tang, Shuo Wang, Yiru Wang, Yu Gao, Yuwen Heng, Zhigang Sun, Zichong Gu","submitted_at":"2025-08-07T06:30:05Z","abstract_excerpt":"Vision-Language-Action (VLA) models have demonstrated potential in autonomous driving. However, two critical challenges hinder their development: (1) Existing VLA architectures are typically based on imitation learning in open-loop setup which tends to capture the recorded behaviors in the dataset, leading to suboptimal and constrained performance, (2) Close-loop training relies heavily on high-fidelity sensor simulation, where domain gaps and computational inefficiencies pose significant barriers. In this paper, we introduce IRL-VLA, a novel close-loop Reinforcement Learning via \\textbf{I}nve"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2508.06571","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-08-07T06:30:05Z","cross_cats_sorted":["cs.CV","cs.RO"],"title_canon_sha256":"ff635ba172f86b5029f93dbc2c10677fcb8b9d4b9559ff96f67bc223742b108b","abstract_canon_sha256":"81692509445b93349d972e9bd287c54acc3e0b08cb85d1eeb06aa6caafc1145c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:54:13.449455Z","signature_b64":"MzX/QIgM6G5/kcB72Q+Ba6cpNSgN8BJZpyrCIRufe1FD+3GQtknjoPjbUDY77phlN/RhyndZjuw3dLnYj9x5AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e7db982554bb6bf9a4c86897f87900a65f1df2efe5b814e61e62edc1a10bb6de","last_reissued_at":"2026-07-05T11:54:13.448970Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:54:13.448970Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"IRL-VLA: Training an Vision-Language-Action Policy via Reward World Model","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","cs.RO"],"primary_cat":"cs.AI","authors_text":"Anqing Jiang, Hao Jiang, Hao Sun, Jijun Wang, Jinhao Chai, Lijuan Zhu, Li Sun, Shichen Tang, Shuo Wang, Yiru Wang, Yu Gao, Yuwen Heng, Zhigang Sun, Zichong Gu","submitted_at":"2025-08-07T06:30:05Z","abstract_excerpt":"Vision-Language-Action (VLA) models have demonstrated potential in autonomous driving. However, two critical challenges hinder their development: (1) Existing VLA architectures are typically based on imitation learning in open-loop setup which tends to capture the recorded behaviors in the dataset, leading to suboptimal and constrained performance, (2) Close-loop training relies heavily on high-fidelity sensor simulation, where domain gaps and computational inefficiencies pose significant barriers. In this paper, we introduce IRL-VLA, a novel close-loop Reinforcement Learning via \\textbf{I}nve"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.06571","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2508.06571/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2508.06571","created_at":"2026-07-05T11:54:13.449029+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.06571v3","created_at":"2026-07-05T11:54:13.449029+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.06571","created_at":"2026-07-05T11:54:13.449029+00:00"},{"alias_kind":"pith_short_12","alias_value":"47NZQJKUXNV7","created_at":"2026-07-05T11:54:13.449029+00:00"},{"alias_kind":"pith_short_16","alias_value":"47NZQJKUXNV7TJGI","created_at":"2026-07-05T11:54:13.449029+00:00"},{"alias_kind":"pith_short_8","alias_value":"47NZQJKU","created_at":"2026-07-05T11:54:13.449029+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":17,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2607.08375","citing_title":"WCog-VLA: A Dual-Level World-Cognitive Vision-Language-Action Model for End-to-End Autonomous Driving","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2607.08072","citing_title":"Post-Training in End-to-End Autonomous Driving","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2606.19836","citing_title":"World Engine: Towards the Era of Post-Training for Autonomous Driving","ref_index":101,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05645","citing_title":"Discrete-WAM: Unified Discrete Vision-Action Token Editing for World-Policy Learning","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2605.31476","citing_title":"IDOL: Inverse-Dynamics-Guided Future Prediction for End-to-End Autonomous Driving","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00113","citing_title":"World Models for Robotic Manipulation: A Survey","ref_index":106,"is_internal_anchor":false},{"citing_arxiv_id":"2605.31116","citing_title":"NTR: Neural Token Reconstruction for Scene Token Bottleneck in End-to-End Driving","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21139","citing_title":"Distill to Think, Foresee to Act: Cognitive-Physical Reinforcement Learning for Autonomous Driving","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21139","citing_title":"Distill to Think, Foresee to Act: Cognitive-Physical Reinforcement Learning for Autonomous Driving","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2509.24948","citing_title":"World-Env: Leveraging World Model as a Virtual Environment for VLA Post-Training","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2511.00088","citing_title":"Alpamayo-R1: Bridging Reasoning and Action Prediction for Generalizable Autonomous Driving in the Long Tail","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2511.18960","citing_title":"AVA-VLA: Improving Vision-Language-Action models with Active Visual Attention","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2512.10226","citing_title":"Latent Chain-of-Thought World Modeling for End-to-End Driving","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08975","citing_title":"Latency Analysis and Optimization of Alpamayo 1 via Efficient Trajectory Generation","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19710","citing_title":"SpanVLA: Efficient Action Bridging and Learning from Negative-Recovery Samples for Vision-Language-Action Model","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2604.13654","citing_title":"Vision-and-Language Navigation for UAVs: Progress, Challenges, and a Research Roadmap","ref_index":169,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16592","citing_title":"Human Cognition in Machines: A Unified Perspective of World Models","ref_index":80,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/47NZQJKUXNV7TJGINCL7Q6IAUZ","json":"https://pith.science/pith/47NZQJKUXNV7TJGINCL7Q6IAUZ.json","graph_json":"https://pith.science/api/pith-number/47NZQJKUXNV7TJGINCL7Q6IAUZ/graph.json","events_json":"https://pith.science/api/pith-number/47NZQJKUXNV7TJGINCL7Q6IAUZ/events.json","paper":"https://pith.science/paper/47NZQJKU"},"agent_actions":{"view_html":"https://pith.science/pith/47NZQJKUXNV7TJGINCL7Q6IAUZ","download_json":"https://pith.science/pith/47NZQJKUXNV7TJGINCL7Q6IAUZ.json","view_paper":"https://pith.science/paper/47NZQJKU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.06571&json=true","fetch_graph":"https://pith.science/api/pith-number/47NZQJKUXNV7TJGINCL7Q6IAUZ/graph.json","fetch_events":"https://pith.science/api/pith-number/47NZQJKUXNV7TJGINCL7Q6IAUZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/47NZQJKUXNV7TJGINCL7Q6IAUZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/47NZQJKUXNV7TJGINCL7Q6IAUZ/action/storage_attestation","attest_author":"https://pith.science/pith/47NZQJKUXNV7TJGINCL7Q6IAUZ/action/author_attestation","sign_citation":"https://pith.science/pith/47NZQJKUXNV7TJGINCL7Q6IAUZ/action/citation_signature","submit_replication":"https://pith.science/pith/47NZQJKUXNV7TJGINCL7Q6IAUZ/action/replication_record"}},"created_at":"2026-07-05T11:54:13.449029+00:00","updated_at":"2026-07-05T11:54:13.449029+00:00"}