Metraipolis › archive › dc_720dbd7a7ad5

RAF-VLA: Representation Alignment with the Future for End-to-End Autonomous Driving — arXiv:2609.17728

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.17728
License (as archived)
CC0-1.0
Archived
2026-10-15T16:55:35.426Z
SHA-256 of the archived text
8b0d275128ea27ad3c0d4a7897a5e2cd18e206075fd213558df14d29732de1d2
Stable URL of this page
https://metraipolis.com/documento/dc_720dbd7a7ad5

How to cite

RAF-VLA: Representation Alignment with the Future for End-to-End Autonomous Driving — arXiv:2609.17728 — arXiv (en), https://arxiv.org/abs/2609.17728 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_720dbd7a7ad5 (Metraipolis archive dc_720dbd7a7ad5, sha256 8b0d275128ea27ad3c0d4a7897a5e2cd18e206075fd213558df14d29732de1d2)
BibTeX
@misc{dc720dbd7a7ad5,
  title = {RAF-VLA: Representation Alignment with the Future for End-to-End Autonomous Driving — arXiv:2609.17728},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.17728},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_720dbd7a7ad5; sha256 8b0d275128ea27ad3c0d4a7897a5e2cd18e206075fd213558df14d29732de1d2},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.17728. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Recent Vision-Language-Action (VLA) models for autonomous driving have incorporated world modeling by predicting future driving scenes alongside driving actions, demonstrating strong planning performance. Future driving scenes are utilized as dense supervision, encouraging the policy to learn rich internal representations useful for planning. However, these World-Modeling VLAs rely on explicit future generation to learn such representations, thereby introducing two key limitations: additional training burden and inference latency. To address these limitations, we propose RAF-VLA (Representation Alignment with the Future), a VLA-based autonomous driving framework that shapes planning-relevant internal representations through direct guidance from future-frame representations. RAF-VLA employs Future-Aligned Supervised Fine-Tuning, in which a straightforward regularization aligns the policy's hidden states with future-frame representations obtained from a pretrained world encoder while learning driving actions. This simple alignment allows RAF-VLA to avoid the training burden and inference latency associated with future generation. Extensive experiments on the NAVSIM benchmark show that RAF-VLA achieves competitive planning performance against state-of-the-art VLA planners with substantially fewer training samples seen. Moreover, RAF-VLA incurs only 3.8% training overhead and a negligible 1 ms inference overhead.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-17 in cs.RO).
Categorie arXiv: Robotics (cs.RO).
Pagina del lavoro: https://arxiv.org/abs/2609.17728 — 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 2026-10-15T16:13:27.232Z dalla pagina https://arxiv.org/catchup/cs.RO/2026-09-17?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).