Metraipolis › archive › dc_701858aa357a
Let LiDAR Decide Where to Look: Geometry-Measured Token Selection for Efficient Sparse Latent World Models — arXiv:2610.40935
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.40935
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T05:46:11.617Z
- SHA-256 of the archived text
1f8e07fd7605f7a185784e7870842f4c7f22d7a6098ab67390678b7cad0e2a00- Stable URL of this page
https://metraipolis.com/documento/dc_701858aa357a
How to cite
Let LiDAR Decide Where to Look: Geometry-Measured Token Selection for Efficient Sparse Latent World Models — arXiv:2610.40935 — arXiv (en), https://arxiv.org/abs/2610.40935 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_701858aa357a (Metraipolis archive dc_701858aa357a, sha256 1f8e07fd7605f7a185784e7870842f4c7f22d7a6098ab67390678b7cad0e2a00)
BibTeX
@misc{dc701858aa357a,
title = {Let LiDAR Decide Where to Look: Geometry-Measured Token Selection for Efficient Sparse Latent World Models — arXiv:2610.40935},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.40935},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_701858aa357a; sha256 1f8e07fd7605f7a185784e7870842f4c7f22d7a6098ab67390678b7cad0e2a00},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.40935. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Vision-centric autonomous driving models encode thousands of image tokens per frame, yet most depict sky, buildings, or empty road---content irrelevant to planning. Worse, these tokens dilute the planner's attention: we show that randomly discarding 75% of tokens already outperforms the dense model, because the planner can concentrate on the remaining driving-critical content. Building on this finding, we present Geometry-Aware Token Selection (GATS), which uses LiDAR and radar point projections to select a small fraction of image tokens before the vision backbone runs. Because the selector operates on measured 3D geometry rather than visual features, the backbone encodes only the selected patches, cutting end-to-end FLOPs by up to $18\times$. Each kept token receives a 3D anchor from its LiDAR point and is fed to a self-supervised latent world model that predicts future features and 3D displacements without box annotations. At 25% budget, GATS achieves a 3-sec L2 displacement error of 0.520 m on nuScenes and an EPDMS of 88.9 on NAVSIM---surpassing the dense baseline (0.573 m / 85.9) at $6.6\times$ fewer total FLOPs. At an extreme 10% budget (24 GFLOPs, $18\times$ reduction), it still beats the dense baseline on both nuScenes (0.537 m vs. 0.573 m L2) and NAVSIM (87.9 vs. 85.9 EPDMS). On 739 intent-critical scenarios, GATS covers 95.8% of hazardous vehicles and achieves a 0.09% collision rate---73.5% lower than the dense model (0.34%). Tema degli esploratori: statistica e scienza dei dati (elenco arXiv cs.CV, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2610.40935 — 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 2026-11-03T05:33:32.264Z dalla pagina https://arxiv.org/list/cs.CV/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).