Metraipolis › archive › dc_bf6a750fdbc0
CAR-VLA: Complexity-Aware and Risk-Adaptive Reasoning for Autonomous Driving — arXiv:2609.34387
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.34387
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T23:52:06.495Z
- SHA-256 of the archived text
04492df5f202261a5ad3ab0770260b1dd571bbdbfe2e2da7dc7d4d5825229b0b- Stable URL of this page
https://metraipolis.com/documento/dc_bf6a750fdbc0
How to cite
CAR-VLA: Complexity-Aware and Risk-Adaptive Reasoning for Autonomous Driving — arXiv:2609.34387 — arXiv (en), https://arxiv.org/abs/2609.34387 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bf6a750fdbc0 (Metraipolis archive dc_bf6a750fdbc0, sha256 04492df5f202261a5ad3ab0770260b1dd571bbdbfe2e2da7dc7d4d5825229b0b)
BibTeX
@misc{dcbf6a750fdbc0,
title = {CAR-VLA: Complexity-Aware and Risk-Adaptive Reasoning for Autonomous Driving — arXiv:2609.34387},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.34387},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bf6a750fdbc0; sha256 04492df5f202261a5ad3ab0770260b1dd571bbdbfe2e2da7dc7d4d5825229b0b},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34387. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Existing adaptive reasoning methods for driving Vision-Language-Action (VLA) models primarily focus on whether to reason, overlooking how reasoning should differ across driving situations. Our key insight is that while scene complexity informs reasoning depth, dynamic risk is equally critical for deciding how to reason in time-critical situations. We therefore propose CAR-VLA, a unified driving VLA model that jointly considers scene complexity and dynamic risk to guide reasoning depth, urgency, and focus. CAR-VLA maps four complexity--risk categories to three reasoning modes: \textit{Fast Intuition} for direct trajectory generation in simple low-risk scenes, \textit{Slow Thinking} for deliberate reasoning in complex low-risk scenes, and \textit{Reflex Response} for compact, hazard-focused reasoning in high-risk scenes regardless of complexity. Rather than merely shortening deliberation, Reflex Response centers reasoning on the most critical hazard and the immediate safe response. We train CAR-VLA through progressive supervised learning that links scene assessment, reasoning-mode selection, and trajectory generation, followed by reasoning-augmented reinforcement learning to improve driving quality and reasoning behavior. Experiments on NAVSIM v1(91.1 PDMS), NAVSIM v2(90.3 EPDMS), and Navhard(35.0 EPDMS) demonstrate competitive driving performance. Qualitative comparisons on navtest and in-house high-risk scenarios further illustrate risk-aware reasoning and hazard-responsive trajectory generation. The code for this paper will be released publicly at: this https URL
Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-29 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.34387 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T23:29:27.362Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).