Pith. sign in

REVIEW 87 references

LiAuto-GeoX: Efficient Grounded Driving Transformer

T0 review · reviewed 2026-06-28 · grok-4.3

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A 155M-parameter transformer distilled from a large LiDAR-grounded geometry model runs at 220 FPS for real-time dense 3D reconstruction in driving scenes.

desk verdict The paper's distillation losses for a 155M driving geometry model are the concrete new element, but the abstract gives no experimental details to back the 220 FPS and transfer claims. read the letter →

arxiv 2606.05774 v1 pith:Q2W32UDU submitted 2026-06-04 cs.CV

Jiawei Lian , Haoyi Sun , Yang Wu , Lifu Mu , Siyuan Wang , Le Hui , Ning Mao , Tao Wei
show 3 more authors
Pan Zhou Kun Zhan Jian Yang
This is my paper
classification cs.CV
keywords dense3Dreconstructionautonomousdrivingknowledgedistillationtransformermodelreal-timeinferencegeometricgroundingsurround-viewcamerasoccupancyprediction
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper introduces LiAuto-GeoX as a compact onboard model for ego-centric 3D scene understanding that starts from a high-capacity model trained on large-scale surround-view data with sparse LiDAR priors for grounding. It then applies a geometry-preserving distillation process using mask-guided depth-aware distillation to keep fine metric details and relative-pose relational distillation to maintain cross-view consistency. This yields a deployable model that operates at 220 FPS on KITTI while preserving reconstruction quality and supplying geometric features that improve downstream tasks such as trajectory prediction to 90.6 PDMS, occupancy prediction to 24.63 mIoU, and future-frame prediction to 47.67 IoU.

What carries the argument

Geometry-preserving distillation framework that combines mask-guided depth-aware distillation to retain metric structures and relative-pose relational distillation to enforce cross-view consistency.

What would settle it

A direct comparison on a held-out driving dataset showing that the 155M model loses long-range depth accuracy or cross-view consistency relative to the teacher model would falsify the claim that the distillation framework preserves the required geometric fidelity.

Watch

Extended reading notes

Core claim

LiAuto-GeoX learns a high-capacity driving geometry model from large-scale surround-view data with sparse LiDAR priors, then distills the capability into a 155M-parameter compact model through mask-guided depth-aware distillation that emphasizes geometrically informative regions and relative-pose relational distillation that enforces cross-view spatial consistency, resulting in 220 FPS dense reconstruction on KITTI that transfers directly to autonomy tasks with the reported performance numbers.

Load-bearing premise

The mask-guided depth-aware and relative-pose relational distillation steps successfully transfer fine-grained metric structures and cross-view spatial consistency from the large teacher model to the compact student.

Editorial extensions

If this is right

  • The distilled model supports real-time onboard dense 3D reconstruction at 220 FPS on KITTI.
  • The retained geometry improves trajectory prediction to 90.6 PDMS.
  • The retained geometry improves occupancy prediction to 24.63 mIoU.
  • The retained geometry improves future-frame prediction to 47.67 IoU.
  • Efficient dense reconstruction can serve as a foundational representation for multiple autonomy tasks rather than only a perception output.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same distillation recipe could be tested on other surround-view camera rigs or additional sensor suites to check transferability.
  • The geometry features might be fused into end-to-end planning networks to measure gains in closed-loop driving metrics.
  • If LiDAR is unavailable at deployment, the model could be further adapted with self-supervised signals derived only from video.
  • Scaling the teacher model size or training data volume would likely produce even stronger compact students under the same distillation losses.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

0 major / 0 minor

Summary. The manuscript presents LiAuto-GeoX, an efficient grounded driving transformer for deployable ego-centric 3D scene understanding in autonomous driving. The approach involves training a high-capacity teacher model from large-scale surround-view data with sparse LiDAR priors for geometric grounding, followed by distillation to a 155M-parameter student model using mask-guided depth-aware distillation and relative-pose relational distillation. The student model is claimed to run at 220 FPS on KITTI while maintaining high-fidelity dense reconstruction and achieving strong transfer performance on downstream tasks including 90.6 PDMS in trajectory prediction, 24.63 mIoU in occupancy prediction, and 47.67 IoU in future-frame prediction.

Significance. If the results hold, this work is significant for autonomous driving perception by demonstrating that dense 3D reconstruction can be made real-time and onboard-deployable while serving as a transferable geometric foundation for trajectory, occupancy, and future-frame prediction tasks. The geometry-preserving distillation approach, combining mask-guided depth-aware and relative-pose relational losses, addresses a practical gap between large visual geometry models and resource-constrained deployment. Concrete metrics on the KITTI benchmark and downstream tasks constitute a strength when accompanied by rigorous ablations and hardware specifications in the full manuscript.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive assessment of LiAuto-GeoX and the recommendation to accept. The recognition that the geometry-preserving distillation approach addresses a practical gap in real-time onboard deployment is appreciated.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; claims are empirical

full rationale

The provided abstract and context contain no equations, derivations, or self-citations that reduce predictions to fitted inputs by construction. Performance numbers (220 FPS, PDMS/mIoU/IoU scores) are presented as measured outcomes of training and distillation on data, not as outputs forced by the method's own definitions. The distillation framework is described at a high level without algebraic self-reference. This is the common case of a self-contained empirical pipeline with no load-bearing circular steps.

Assumptions & free parameters 0 free parameters · 0 assumptions · 0 invented entities

The abstract does not provide sufficient detail to identify any free parameters, axioms, or invented entities used in the work.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LiAuto-GeoX: Efficient Grounded Driving Transformer." pith.science (2026). https://pith.science/paper/Q2W32UDU

@misc{pith2026260605774,
  author       = {Pith},
  title        = {Pith review of: LiAuto-GeoX: Efficient Grounded Driving Transformer},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Q2W32UDU}},
  note         = {Machine review of arXiv:2606.05774}
}
read the original abstract

Dense 3D reconstruction has demonstrated immense potential for spatial understanding, yet its viability as a real-time, onboard representation for autonomous driving remains an open challenge. Existing large-scale visual geometry models typically require substantial computational resources and lack the long-range geometric fidelity, surround-view consistency, and real-time efficiency demanded by dynamic driving environments. To bridge this gap, we present \textbf{LiAuto-GeoX}, an efficient grounded driving transformer designed for deployable, ego-centric 3D scene understanding. Our approach begins by learning a high-capacity driving geometry model from large-scale surround-view data, utilizing sparse LiDAR priors to provide robust geometric grounding in distant, ambiguous, or structure-sparse regions. We then instantiate this capability into a highly compact 155M-parameter onboard model through a novel geometry-preserving distillation framework. This framework employs mask-guided depth-aware distillation to retain fine-grained metric structures by emphasizing geometrically informative regions, and relative-pose relational distillation to enforce cross-view spatial consistency through pose-induced geometric relations. Extensive evaluations reveal that \textbf{LiAuto-GeoX} runs at 220 FPS on KITTI while maintaining high-fidelity dense reconstruction, enabling real-time deployment. The learned geometry transfers seamlessly to downstream autonomy tasks, achieving 90.6 PDMS in trajectory prediction, 24.63 mIoU in occupancy prediction, and 47.67 IoU in future-frame prediction. These all demonstrate that efficient dense 3D reconstruction can transcend its traditional role as a perception target to serve as a scalable, foundational geometric representation for next-generation autonomous driving.

Figures

Figures reproduced from arXiv: 2606.05774 by the authors.

Figure 1
Figure 1. Overview of LiAuto-GeoX. LiAuto-GeoX provides an efficient driving geometry model for surround-view 3D reconstruction and downstream autonomy tasks. It strikes a compact yet effective accuracy–efficiency balance, accommodates diverse camera configurations, and transfers the learned dense geometry to tasks such as pose estimation, depth estimation, 3D reconstruction, trajectory prediction, occupancy prediction, and f… view at source ↗
Figure 2
Figure 2. Overall pipeline of LiAuto-GeoX. LiAuto-GeoX first trains a high-capacity teacher model to learn dense driving geometry from calibrated multi-view RGB inputs, then distills its geometric capability into a compact student model with task supervision, token-mask condition, and relational constraints. During inference, only the student is deployed to produce dense 3D reconstructions under flexible surround-view camera … view at source ↗
Figure 3
Figure 3. Mask-Guided Depth-Aware Distillation. The frozen teacher model extracts high￾activation regions to form a token-level mask, which conditions the student model through a lightweight depth-mask adapter. This training-time guidance focuses the compact student on geomet￾rically informative regions without adding inference cost. where z T i is the i-th teacher token and LN(·) denotes layer normalization. A binary activat… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Relative-Pose Relational Distillation. We normalize the teacher and student camera￾conditioned tokens. We then construct pairwise relational matrices across input views to distill cross￾view geometric relations. The student is supervised with an off-diagonal relational…
Figure 5
Figure 5. Figure 5: Qualitative comparison of multi-view 3D reconstruction. We compare our method against baselines (VGGT, OmniVGGT, PI3, DVGT) using their default open-source configurations. Each result displays the Ego View (top) and Top-Down View (bottom). Across varying camera counts …
Figure 6
Figure 6. Figure 6: Qualitative comparison of surround-view reconstruction. We compare LiAuto-GeoX with DVGT on Waymo and DDAD sequences. Red and blue trajectories denote DVGT and our method, respectively. LiAuto-GeoX reconstructs cleaner road boundaries, more complete structures, and mor…
Figure 7
Figure 7. Figure 7: Qualitative visualizations. Surround-view inputs and reconstructed 3D driving scenes with driving trajectories. Using tokens with activations above the average achieves the best overall performance, reaching 0.614 Acc, 1.871 Comp, 0.638 NC, 0.075 Abs Rel, and 0.914 (δ …

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

87 extracted references · 1 canonical work pages

  1. [1]

    nuscenes: A multimodal dataset for autonomous driving

    Holger Caesar, Varun Bankiti, Alex H Lang, Sourabh V ora, Venice Erin Liong, Qiang Xu, Anush Krishnan, Yu Pan, Giancarlo Baldan, and Oscar Beijbom. nuscenes: A multimodal dataset for autonomous driving. InCVPR, pages 11621–11631, 2020

    2020

  2. [2]

    Vadv2: End-to-end vectorized autonomous driving via probabilistic planning.ICLR, 2026

    Shaoyu Chen, Bo Jiang, Hao Gao, Bencheng Liao, Qing Xu, Qian Zhang, Chang Huang, Wenyu Liu, and Xinggang Wang. Vadv2: End-to-end vectorized autonomous driving via probabilistic planning.ICLR, 2026

    2026

  3. [3]

    Drivinggpt: Unifying driving world modeling and planning with multi-modal autoregressive transformers

    Yuntao Chen, Yuqi Wang, and Zhaoxiang Zhang. Drivinggpt: Unifying driving world modeling and planning with multi-modal autoregressive transformers. InICCV, pages 26890–26900, 2025

    2025

  4. [4]

    Bevdistill: Cross-modal bev distillation for multi-view 3d object detection.arXiv, 2022

    Zehui Chen, Zhenyu Li, Shiquan Zhang, Liangji Fang, Qinhong Jiang, and Feng Zhao. Bevdistill: Cross-modal bev distillation for multi-view 3d object detection.arXiv, 2022

    2022

  5. [5]

    Openscene: The largest up-to-date 3d occupancy prediction bench- mark in autonomous driving

    OpenScene Contributors. Openscene: The largest up-to-date 3d occupancy prediction bench- mark in autonomous driving. InCVPR, pages 18–22, 2023

    2023

  6. [6]

    Sparseworld: A flexible, adaptive, and efficient 4d occupancy world model powered by sparse and dynamic queries

    Chenxu Dang, Haiyan Liu, Jason Bao, Pei An, Xinyue Tang, An Pan, Jie Ma, Bingchuan Sun, and Yan Wang. Sparseworld: A flexible, adaptive, and efficient 4d occupancy world model powered by sparse and dynamic queries. InAAAI, volume 40, pages 3497–3505, 2026

    2026

  7. [7]

    Navsim: Data-driven non-reactive autonomous vehicle simulation and benchmarking.NeurIPS, 37:28706–28719, 2024

    Daniel Dauner, Marcel Hallgarten, Tianyu Li, Xinshuo Weng, Zhiyu Huang, Zetong Yang, Hongyang Li, Igor Gilitschenski, Boris Ivanovic, Marco Pavone, et al. Navsim: Data-driven non-reactive autonomous vehicle simulation and benchmarking.NeurIPS, 37:28706–28719, 2024

    2024

  8. [8]

    Sparseworld-tc: Trajectory-conditioned sparse occupancy world model.arXiv, 2025

    Jiayuan Du, Yiming Zhao, Zhenglong Guo, Yong Pan, Wenbo Hou, Zhihui Hao, Kun Zhan, and Qijun Chen. Sparseworld-tc: Trajectory-conditioned sparse occupancy world model.arXiv, 2025

    2025

Show all 87 references
  1. [9]

    Artemis: Autoregressive end-to-end trajectory planning with mixture of experts for autonomous driving.IEEE RAL, 11(1):226–233, 2025

    Renju Feng, Ning Xi, Duanfeng Chu, Rukang Wang, Zejian Deng, Anzheng Wang, Liping Lu, Jinxiang Wang, and Yanjun Huang. Artemis: Autoregressive end-to-end trajectory planning with mixture of experts for autonomous driving.IEEE RAL, 11(1):226–233, 2025

    2025

  2. [10]

    Are we ready for autonomous driving? the kitti vision benchmark suite

    Andreas Geiger, Philip Lenz, and Raquel Urtasun. Are we ready for autonomous driving? the kitti vision benchmark suite. InCVPR, pages 3354–3361. IEEE, 2012

    2012

  3. [11]

    Dome: Taming diffusion model into high-fidelity controllable occupancy world model.arXiv, 2024

    Songen Gu, Wei Yin, Bu Jin, Xiaoyang Guo, Junming Wang, Haodong Li, Qian Zhang, and Xiaoxiao Long. Dome: Taming diffusion model into high-fidelity controllable occupancy world model.arXiv, 2024

    2024

  4. [12]

    3d packing for self-supervised monocular depth estimation

    Vitor Guizilini, Rares Ambrus, Sudeep Pillai, Allan Raventos, and Adrien Gaidon. 3d packing for self-supervised monocular depth estimation. InCVPR, pages 2485–2494, 2020

    2020

  5. [13]

    Distilling the knowledge in a neural network

    Geoffrey Hinton, Oriol Vinyals, and Jeff Dean. Distilling the knowledge in a neural network. arXiv, 2015

    2015

  6. [14]

    Lrm: Large reconstruction model for single image to 3d

    Yicong Hong, Kai Zhang, Jiuxiang Gu, Sai Bi, Yang Zhou, Difan Liu, Feng Liu, Kalyan Sunkavalli, Trung Bui, and Hao Tan. Lrm: Large reconstruction model for single image to 3d. InICLR, volume 2024, pages 50678–50702, 2024

    2024

  7. [15]

    One thousand and one hours: Self-driving motion prediction dataset

    John Houston, Guido Zuidhof, Luca Bergamini, Yawei Ye, Long Chen, Ashesh Jain, Sammy Omari, Vladimir Iglovikov, and Peter Ondruska. One thousand and one hours: Self-driving motion prediction dataset. InCoRL, pages 409–418. PMLR, 2021

    2021

  8. [16]

    Planning-oriented autonomous driving

    Yihan Hu, Jiazhi Yang, Li Chen, Keyu Li, Chonghao Sima, Xizhou Zhu, Siqi Chai, Senyao Du, Tianwei Lin, Wenhai Wang, et al. Planning-oriented autonomous driving. InCVPR, pages 17853–17862, 2023

    2023

  9. [17]

    Bevdet4d: Exploit temporal cues in multi-camera 3d object detection.arXiv, 2022

    Junjie Huang and Guan Huang. Bevdet4d: Exploit temporal cues in multi-camera 3d object detection.arXiv, 2022. 17

    2022

  10. [18]

    Bevdet: High-performance multi-camera 3d object detection in bird-eye-view.arXiv, 2021

    Junjie Huang, Guan Huang, Zheng Zhu, Yun Ye, and Dalong Du. Bevdet: High-performance multi-camera 3d object detection in bird-eye-view.arXiv, 2021

    2021

  11. [19]

    Tig-bev: Multi-view bev 3d object detection via target inner-geometry learning.arXiv, 2022

    Peixiang Huang, Li Liu, Renrui Zhang, Song Zhang, Xinli Xu, Baichao Wang, and Guoyi Liu. Tig-bev: Multi-view bev 3d object detection via target inner-geometry learning.arXiv, 2022

    2022

  12. [20]

    Tri-perspective view for vision-based 3d semantic occupancy prediction

    Yuanhui Huang, Wenzhao Zheng, Yunpeng Zhang, Jie Zhou, and Jiwen Lu. Tri-perspective view for vision-based 3d semantic occupancy prediction. InCVPR, pages 9223–9232, 2023

    2023

  13. [21]

    Occtens: 3d occupancy world model via temporal next-scale prediction.IEEE RAL, 2026

    Bu Jin, Songen Gu, Xiaotao Hu, Yupeng Zheng, Xiaoyang Guo, Qian Zhang, Xiaoxiao Long, and Wei Yin. Occtens: 3d occupancy world model via temporal next-scale prediction.IEEE RAL, 2026

    2026

  14. [22]

    3d gaussian splatting for real-time radiance field rendering.ACM Trans

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimkühler, George Drettakis, et al. 3d gaussian splatting for real-time radiance field rendering.ACM Trans. Graph., 42(4):139–1, 2023

    2023

  15. [23]

    X3kd: Knowledge distillation across modal- ities, tasks and stages for multi-camera 3d object detection

    Marvin Klingner, Shubhankar Borse, Varun Ravi Kumar, Behnaz Rezaei, Venkatraman Narayanan, Senthil Yogamani, and Fatih Porikli. X3kd: Knowledge distillation across modal- ities, tasks and stages for multi-camera 3d object detection. InCVPR, pages 13343–13353, 2023

    2023

  16. [24]

    Grounding image matching in 3d with mast3r

    Vincent Leroy, Yohann Cabon, and Jérôme Revaud. Grounding image matching in 3d with mast3r. InECCV, pages 71–91. Springer, 2024

    2024

  17. [25]

    Semi-supervised vision-centric 3d occupancy world model for autonomous driving

    Xiang Li, Pengfei Li, Yupeng Zheng, Wei Sun, Yan Wang, et al. Semi-supervised vision-centric 3d occupancy world model for autonomous driving. InICLR, volume 2025, pages 62563–62580, 2025

    2025

  18. [26]

    V oxformer: Sparse voxel transformer for camera-based 3d semantic scene completion

    Yiming Li, Zhiding Yu, Christopher Choy, Chaowei Xiao, Jose M Alvarez, Sanja Fidler, Chen Feng, and Anima Anandkumar. V oxformer: Sparse voxel transformer for camera-based 3d semantic scene completion. InCVPR, pages 9087–9098, 2023

    2023

  19. [27]

    Enhancing end-to-end autonomous driving with latent world model.ICLR, 2025

    Yingyan Li, Lue Fan, Jiawei He, Yuqi Wang, Yuntao Chen, Zhaoxiang Zhang, and Tieniu Tan. Enhancing end-to-end autonomous driving with latent world model.ICLR, 2025

    2025

  20. [28]

    End-to-end driving with online trajectory evaluation via bev world model

    Yingyan Li, Yuqi Wang, Yang Liu, Jiawei He, Lue Fan, and Zhaoxiang Zhang. End-to-end driving with online trajectory evaluation via bev world model. InICCV, pages 27137–27146, 2025

    2025

  21. [29]

    Drivevla-w0: World models amplify data scaling law in autonomous driving.ICLR, 2026

    Yingyan Li, Shuyao Shang, Weisong Liu, Bing Zhan, Haochen Wang, Yuqi Wang, Yuntao Chen, Xiaoman Wang, Yasong An, Chufeng Tang, et al. Drivevla-w0: World models amplify data scaling law in autonomous driving.ICLR, 2026

    2026

  22. [30]

    Bevdepth: Acquisition of reliable depth for multi-view 3d object detection

    Yinhao Li, Zheng Ge, Guanyi Yu, Jinrong Yang, Zengran Wang, Yukang Shi, Jianjian Sun, and Zeming Li. Bevdepth: Acquisition of reliable depth for multi-view 3d object detection. InAAAI, volume 37, pages 1477–1485, 2023

    2023

  23. [31]

    Recogdrive: A reinforced cognitive framework for end-to-end autonomous driving.ICLR, 2026

    Yongkang Li, Kaixin Xiong, Xiangyu Guo, Fang Li, Sixu Yan, Gangwei Xu, Lijun Zhou, Long Chen, Haiyang Sun, Bing Wang, et al. Recogdrive: A reinforced cognitive framework for end-to-end autonomous driving.ICLR, 2026

    2026

  24. [32]

    Hydra-mdp: End-to-end multimodal planning with multi-target hydra-distillation.arXiv, 2024

    Zhenxin Li, Kailin Li, Shihao Wang, Shiyi Lan, Zhiding Yu, Yishen Ji, Zhiqi Li, Ziyue Zhu, Jan Kautz, Zuxuan Wu, et al. Hydra-mdp: End-to-end multimodal planning with multi-target hydra-distillation.arXiv, 2024

    2024

  25. [33]

    Bevformer: learning bird’s-eye-view representation from lidar-camera via spatiotemporal transformers.IEEE TPAMI, 47(3):2020–2036, 2024

    Zhiqi Li, Wenhai Wang, Hongyang Li, Enze Xie, Chonghao Sima, Tong Lu, Qiao Yu, and Jifeng Dai. Bevformer: learning bird’s-eye-view representation from lidar-camera via spatiotemporal transformers.IEEE TPAMI, 47(3):2020–2036, 2024

    2020

  26. [34]

    Diffusion-based contextual reconstruction for point cloud segmentation with limited annotations.AAAI, 40(8): 6780–6788, Mar

    Jiawei Lian, Zhengxue Wang, Wentao Qu, Haobo Jiang, Le Hui, and Jian Yang. Diffusion-based contextual reconstruction for point cloud segmentation with limited annotations.AAAI, 40(8): 6780–6788, Mar. 2026. doi: 10.1609/aaai.v40i8.37610. 18

    2026 doi

  27. [35]

    Bevfusion: A simple and robust lidar-camera fusion framework

    Tingting Liang, Hongwei Xie, Kaicheng Yu, Zhongyu Xia, Zhiwei Lin, Yongtao Wang, Tao Tang, Bing Wang, and Zhi Tang. Bevfusion: A simple and robust lidar-camera fusion framework. NeurIPS, 35:10421–10434, 2022

    2022

  28. [36]

    Diffusiondrive: Truncated diffusion model for end-to-end autonomous driving

    Bencheng Liao, Shaoyu Chen, Haoran Yin, Bo Jiang, Cheng Wang, Sixu Yan, Xinbang Zhang, Xiangyu Li, Ying Zhang, Qian Zhang, et al. Diffusiondrive: Truncated diffusion model for end-to-end autonomous driving. InCVPR, pages 12037–12047, 2025

    2025

  29. [37]

    I2- world: Intra-inter tokenization for efficient dynamic 4d scene forecasting

    Zhimin Liao, Ping Wei, Ruijie Zhang, Shuaijia Chen, Haoxuan Wang, and Ziyang Ren. I2- world: Intra-inter tokenization for efficient dynamic 4d scene forecasting. InICCV, pages 25810–25819, 2025

    2025

  30. [38]

    Adathinkdrive: Adaptive thinking via reinforcement learning for autonomous driving.arXiv, 2025

    Yuechen Luo, Fang Li, Shaoqing Xu, Zhiyi Lai, Lei Yang, Qimao Chen, Ziang Luo, Zixun Xie, Shengyin Jiang, Jiaxin Liu, et al. Adathinkdrive: Adaptive thinking via reinforcement learning for autonomous driving.arXiv, 2025

    2025

  31. [39]

    Nerf: Representing scenes as neural radiance fields for view synthesis.CACM, 65 (1):99–106, 2021

    Ben Mildenhall, Pratul P Srinivasan, Matthew Tancik, Jonathan T Barron, Ravi Ramamoorthi, and Ren Ng. Nerf: Representing scenes as neural radiance fields for view synthesis.CACM, 65 (1):99–106, 2021

    2021

  32. [40]

    Mast3r-slam: Real-time dense slam with 3d reconstruction priors

    Riku Murai, Eric Dexheimer, and Andrew J Davison. Mast3r-slam: Real-time dense slam with 3d reconstruction priors. InCVPR, pages 16695–16705, 2025

    2025

  33. [41]

    Dinov2: Learning robust visual features without supervision.arXiv, 2023

    Maxime Oquab, Timothée Darcet, Théo Moutakanni, Huy V o, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, et al. Dinov2: Learning robust visual features without supervision.arXiv, 2023

    2023

  34. [42]

    Renderocc: Vision-centric 3d occupancy prediction with 2d rendering supervision

    Mingjie Pan, Jiaming Liu, Renrui Zhang, Peixiang Huang, Xiaoqi Li, Hongwei Xie, Bing Wang, Li Liu, and Shanghang Zhang. Renderocc: Vision-centric 3d occupancy prediction with 2d rendering supervision. InICRA, pages 12404–12411. IEEE, 2024

    2024

  35. [43]

    Omnivggt: Omni-modality driven visual geometry grounded transformer.arXiv, 2025

    Haosong Peng, Hao Li, Yalun Dai, Yushi Lan, Yihang Luo, Tianyu Qi, Zhengshen Zhang, Yufeng Zhan, Junfei Zhang, Wenchao Xu, et al. Omnivggt: Omni-modality driven visual geometry grounded transformer.arXiv, 2025

    2025

  36. [44]

    Lift, splat, shoot: Encoding images from arbitrary camera rigs by implicitly unprojecting to 3d

    Jonah Philion and Sanja Fidler. Lift, splat, shoot: Encoding images from arbitrary camera rigs by implicitly unprojecting to 3d. InECCV, pages 194–210. Springer, 2020

    2020

  37. [45]

    Multi-modal fusion transformer for end-to-end autonomous driving

    Aditya Prakash, Kashyap Chitta, and Andreas Geiger. Multi-modal fusion transformer for end-to-end autonomous driving. InCVPR, pages 7077–7087, 2021

    2021

  38. [46]

    Fastvggt: Training-free acceleration of visual geometry transformer.arXiv, 2025

    You Shen, Zhipeng Zhang, Yansong Qu, Xiawu Zheng, Jiayi Ji, Shengchuan Zhang, and Liujuan Cao. Fastvggt: Training-free acceleration of visual geometry transformer.arXiv, 2025

    2025

  39. [47]

    Come: Adding scene-centric forecasting control to occupancy world model.NeurIPS, 38:70–96, 2026

    Yining Shi, Kun Jiang, Qiang Meng, Ke Wang, Jiabao Wang, Wenchao Sun, Tuopu Wen, et al. Come: Adding scene-centric forecasting control to occupancy world model.NeurIPS, 38:70–96, 2026

    2026

  40. [48]

    Litevggt: Boosting vanilla vggt via geometry-aware cached token merging

    Zhijian Shu, Cheng Lin, Tao Xie, Wei Yin, Ben Li, Zhiyuan Pu, Weize Li, Yao Yao, Xun Cao, Xiaoyang Guo, et al. Litevggt: Boosting vanilla vggt via geometry-aware cached token merging. arXiv, 2025

    2025

  41. [49]

    Drivelm: Driving with graph visual question answering

    Chonghao Sima, Katrin Renz, Kashyap Chitta, Li Chen, Hanxue Zhang, Chengen Xie, Jens Beißwenger, Ping Luo, Andreas Geiger, and Hongyang Li. Drivelm: Driving with graph visual question answering. InECCV, pages 256–274. Springer, 2024

    2024

  42. [50]

    Training very deep networks

    Rupesh K Srivastava, Klaus Greff, and Jürgen Schmidhuber. Training very deep networks. NeurIPS, 28, 2015

    2015

  43. [51]

    Scalability in perception for autonomous driving: Waymo open dataset

    Pei Sun, Henrik Kretzschmar, Xerxes Dotiwalla, Aurelien Chouard, Vijaysai Patnaik, Paul Tsui, James Guo, Yin Zhou, Yuning Chai, Benjamin Caine, et al. Scalability in perception for autonomous driving: Waymo open dataset. InCVPR, pages 2446–2454, 2020. 19

    2020

  44. [52]

    Occ3d: A large-scale 3d occupancy prediction benchmark for autonomous driving.NeurIPS, 36:64318–64330, 2023

    Xiaoyu Tian, Tao Jiang, Longfei Yun, Yucheng Mao, Huitong Yang, Yue Wang, Yilun Wang, and Hang Zhao. Occ3d: A large-scale 3d occupancy prediction benchmark for autonomous driving.NeurIPS, 36:64318–64330, 2023

    2023

  45. [53]

    Drivevlm: The convergence of autonomous driving and large vision-language models.arXiv, 2024

    Xiaoyu Tian, Junru Gu, Bailin Li, Yicheng Liu, Yang Wang, Zhiyong Zhao, Kun Zhan, Peng Jia, Xianpeng Lang, and Hang Zhao. Drivevlm: The convergence of autonomous driving and large vision-language models.arXiv, 2024

    2024

  46. [54]

    Least-squares estimation of transformation parameters between two point patterns.IEEE TPAMI, 13(4):376–380, 1991

    Shinji Umeyama. Least-squares estimation of transformation parameters between two point patterns.IEEE TPAMI, 13(4):376–380, 1991

    1991

  47. [55]

    Vggt: Visual geometry grounded transformer

    Jianyuan Wang, Minghao Chen, Nikita Karaev, Andrea Vedaldi, Christian Rupprecht, and David Novotny. Vggt: Visual geometry grounded transformer. InCVPR, pages 5294–5306, 2025

    2025

  48. [56]

    Regularizing nighttime weirdness: Efficient self-supervised monocular depth estimation in the dark

    Kun Wang, Zhenyu Zhang, Zhiqiang Yan, Xiang Li, Baobei Xu, Jun Li, and Jian Yang. Regularizing nighttime weirdness: Efficient self-supervised monocular depth estimation in the dark. InICCV, pages 16055–16064, 2021

    2021

  49. [57]

    Continuous 3d perception model with persistent state

    Qianqian Wang, Yifei Zhang, Aleksander Holynski, Alexei A Efros, and Angjoo Kanazawa. Continuous 3d perception model with persistent state. InCVPR, pages 10510–10522, 2025

    2025

  50. [58]

    Dust3r: Geometric 3d vision made easy

    Shuzhe Wang, Vincent Leroy, Yohann Cabon, Boris Chidlovskii, and Jerome Revaud. Dust3r: Geometric 3d vision made easy. InCVPR, pages 20697–20709, 2024

    2024

  51. [59]

    Openoccupancy: A large scale benchmark for surrounding semantic occupancy perception

    Xiaofeng Wang, Zheng Zhu, Wenbo Xu, Yunpeng Zhang, Yi Wei, Xu Chi, Yun Ye, Dalong Du, Jiwen Lu, and Xingang Wang. Openoccupancy: A large scale benchmark for surrounding semantic occupancy perception. InICCV, pages 17850–17859, 2023

    2023

  52. [60]

    pi3: Permutation-equivariant visual geometry learning.arXiv, 2025

    Yifan Wang, Jianjun Zhou, Haoyi Zhu, Wenzheng Chang, Yang Zhou, Zizun Li, Junyi Chen, Jiangmiao Pang, Chunhua Shen, and Tong He. pi3: Permutation-equivariant visual geometry learning.arXiv, 2025

    2025

  53. [61]

    Distillbev: Boosting multi-camera 3d object detection with cross-modal knowledge distillation

    Zeyu Wang, Dingwen Li, Chenxu Luo, Cihang Xie, and Xiaodong Yang. Distillbev: Boosting multi-camera 3d object detection with cross-modal knowledge distillation. InICCV, pages 8637–8646, 2023

    2023

  54. [62]

    Surroundocc: Multi-camera 3d occupancy prediction for autonomous driving

    Yi Wei, Linqing Zhao, Wenzhao Zheng, Zheng Zhu, Jie Zhou, and Jiwen Lu. Surroundocc: Multi-camera 3d occupancy prediction for autonomous driving. InICCV, pages 21729–21740, 2023

    2023

  55. [63]

    Weathergen: A unified diverse weather generator for lidar point clouds via spider mamba diffusion

    Yang Wu, Yun Zhu, Kaihua Zhang, Jianjun Qian, Jin Xie, and Jian Yang. Weathergen: A unified diverse weather generator for lidar point clouds via spider mamba diffusion. InCVPR, pages 17019–17028, 2025

    2025

  56. [64]

    Gem: Generating lidar world model via deformable mamba.arXiv, 2026

    Yang Wu, Zhaojiang Liu, Qiang Meng, Youquan Liu, Renliang Weng, Jianjun Qian, Jian Yang, and Jin Xie. Gem: Generating lidar world model via deformable mamba.arXiv, 2026

    2026

  57. [65]

    Distill to think, foresee to act: Cognitive-physical reinforcement learning for autonomous driving.arXiv, 2026

    Yang Wu, Qiang Meng, Zhaojiang Liu, Youquan Liu, Jian Yang, and Jin Xie. Distill to think, foresee to act: Cognitive-physical reinforcement learning for autonomous driving.arXiv, 2026

    2026

  58. [66]

    Drivelaw: Unifying planning and video generation in a latent driving world.CVPR, 2026

    Tianze Xia, Yongkang Li, Lijun Zhou, Jingfeng Yao, Kaixin Xiong, Haiyang Sun, Bing Wang, Kun Ma, Guang Chen, Hangjun Ye, et al. Drivelaw: Unifying planning and video generation in a latent driving world.CVPR, 2026

    2026

  59. [67]

    Pandaset: Advanced sensor suite dataset for autonomous driving

    Pengchuan Xiao, Zhenlei Shao, Steven Hao, Zishuo Zhang, Xiaolin Chai, Judy Jiao, Zesong Li, Jian Wu, Kai Sun, Kun Jiang, et al. Pandaset: Advanced sensor suite dataset for autonomous driving. InITSC, pages 3095–3101. IEEE, 2021

    2021

  60. [68]

    Goalflow: Goal-driven flow matching for multimodal trajectories generation in end-to-end autonomous driving

    Zebin Xing, Xingyu Zhang, Yang Hu, Bo Jiang, Tong He, Qian Zhang, Xiaoxiao Long, and Wei Yin. Goalflow: Goal-driven flow matching for multimodal trajectories generation in end-to-end autonomous driving. InCVPR, pages 1602–1611, 2025

    2025

  61. [69]

    Delta- triplane transformers as occupancy world models.arXiv, 2025

    Haoran Xu, Peixi Peng, Guang Tan, Yiqian Chang, Yisen Zhao, and Yonghong Tian. Delta- triplane transformers as occupancy world models.arXiv, 2025. 20

    2025

  62. [70]

    Occ-llm: Enhancing autonomous driving with occupancy-based large language models

    Tianshuo Xu, Hao Lu, Xu Yan, Yingjie Cai, Bingbing Liu, and Yingcong Chen. Occ-llm: Enhancing autonomous driving with occupancy-based large language models. InICRA, pages 8434–8441. IEEE, 2025

    2025

  63. [71]

    Depth anything: Unleashing the power of large-scale unlabeled data

    Lihe Yang, Bingyi Kang, Zilong Huang, Xiaogang Xu, Jiashi Feng, and Hengshuang Zhao. Depth anything: Unleashing the power of large-scale unlabeled data. InCVPR, pages 10371– 10381, 2024

    2024

  64. [72]

    Depth anything v2.NeurIPS, 37:21875–21911, 2024

    Lihe Yang, Bingyi Kang, Zilong Huang, Zhen Zhao, Xiaogang Xu, Jiashi Feng, and Hengshuang Zhao. Depth anything v2.NeurIPS, 37:21875–21911, 2024

    2024

  65. [73]

    Masked generative distillation

    Zhendong Yang, Zhe Li, Mingqi Shao, Dachuan Shi, Zehuan Yuan, and Chun Yuan. Masked generative distillation. InECCV, pages 53–69. Springer, 2022

    2022

  66. [74]

    Drivesuprim: Towards precise trajectory selection for end-to-end planning

    Wenhao Yao, Zhenxin Li, Shiyi Lan, Zi Wang, Xinglong Sun, Jose M Alvarez, and Zuxuan Wu. Drivesuprim: Towards precise trajectory selection for end-to-end planning. InAAAI, volume 40, pages 11910–11918, 2026

    2026

  67. [75]

    Mvsnet: Depth inference for unstructured multi-view stereo

    Yao Yao, Zixin Luo, Shiwei Li, Tian Fang, and Long Quan. Mvsnet: Depth inference for unstructured multi-view stereo. InECCV, pages 767–783, 2018

    2018

  68. [76]

    Flashocc: Fast and memory-efficient occupancy prediction via channel-to-height plugin.arXiv, 2023

    Zichen Yu, Changyong Shu, Jiajun Deng, Kangjie Lu, Zongdai Liu, Jiangyong Yu, Dawei Yang, Hui Li, and Yan Chen. Flashocc: Fast and memory-efficient occupancy prediction via channel-to-height plugin.arXiv, 2023

    2023

  69. [77]

    Paying more attention to attention: Improving the performance of convolutional neural networks via attention transfer.arXiv, 2016

    Sergey Zagoruyko and Nikos Komodakis. Paying more attention to attention: Improving the performance of convolutional neural networks via attention transfer.arXiv, 2016

    2016

  70. [78]

    An efficient occupancy world model via decoupled dynamic flow and image-assisted training.arXiv, 2024

    Haiming Zhang, Ying Xue, Xu Yan, Jiacheng Zhang, Weichao Qiu, Dongfeng Bai, Bingbing Liu, Shuguang Cui, and Zhen Li. An efficient occupancy world model via decoupled dynamic flow and image-assisted training.arXiv, 2024

    2024

  71. [79]

    Epona: Autoregressive diffusion world model for autonomous driving

    Kaiwen Zhang, Zhenyu Tang, Xiaotao Hu, Xingang Pan, Xiaoyang Guo, Yuan Liu, Jingwei Huang, Li Yuan, Qian Zhang, Xiao-Xiao Long, et al. Epona: Autoregressive diffusion world model for autonomous driving. InICCV, pages 27220–27230, 2025

    2025

  72. [80]

    Pointdistiller: Structured knowledge distillation towards efficient and compact 3d detection

    Linfeng Zhang, Runpei Dong, Hung-Shuo Tai, and Kaisheng Ma. Pointdistiller: Structured knowledge distillation towards efficient and compact 3d detection. InCVPR, pages 21791– 21801, 2023

    2023

  73. [81]

    Copilot4d: Learning unsupervised world models for autonomous driving via discrete diffusion

    Lunjun Zhang, Yuwen Xiong, Ze Yang, Sergio Casas, Rui Hu, and Raquel Urtasun. Copilot4d: Learning unsupervised world models for autonomous driving via discrete diffusion. InICLR, volume 2024, pages 7269–7299, 2024

    2024

  74. [82]

    Decoupled knowledge distillation

    Borui Zhao, Quan Cui, Renjie Song, Yiyu Qiu, and Jiajun Liang. Decoupled knowledge distillation. InCVPR, pages 11953–11962, 2022

    2022

  75. [83]

    From forecasting to planning: Policy world model for collaborative state-action prediction.NeurIPS, 38:134585–134611, 2026

    Zhida Zhao, Talas Fu, Yifan Wang, Lijun Wang, and Huchuan Lu. From forecasting to planning: Policy world model for collaborative state-action prediction.NeurIPS, 38:134585–134611, 2026

    2026

  76. [84]

    Unidistill: A universal cross-modality knowledge distillation framework for 3d object detection in bird’s-eye view

    Shengchao Zhou, Weizhou Liu, Chen Hu, Shuchang Zhou, and Chao Ma. Unidistill: A universal cross-modality knowledge distillation framework for 3d object detection in bird’s-eye view. In CVPR, pages 5116–5125, 2023

    2023

  77. [85]

    Autovla: A vision-language-action model for end-to-end autonomous driving with adaptive reasoning and reinforcement fine-tuning.NeurIPS, 2025

    Zewei Zhou, Tianhui Cai, Seth Z Zhao, Yun Zhang, Zhiyu Huang, Bolei Zhou, and Jiaqi Ma. Autovla: A vision-language-action model for end-to-end autonomous driving with adaptive reasoning and reinforcement fine-tuning.NeurIPS, 2025

    2025

  78. [86]

    Dvgt: Driving visual geometry transformer.arXiv, 2025

    Sicheng Zuo, Zixun Xie, Wenzhao Zheng, Shaoqing Xu, Fang Li, Shengyin Jiang, Long Chen, Zhi-Xin Yang, and Jiwen Lu. Dvgt: Driving visual geometry transformer.arXiv, 2025

    2025

  79. [87]

    Dvgt-2: Vision-geometry-action model for autonomous driving at scale.arXiv, 2026

    Sicheng Zuo, Zixun Xie, Wenzhao Zheng, Shaoqing Xu, Fang Li, Hanbing Li, Long Chen, Zhi-Xin Yang, and Jiwen Lu. Dvgt-2: Vision-geometry-action model for autonomous driving at scale.arXiv, 2026. 21

    2026

Pith tools

Reviewed June 28, 2026 · model on record in the stance chip above.