Pith. sign in

REVIEW 3 major objections 4 minor 300 references

This survey claims BEV perception becomes safer as it moves from single-sensor, to fused multi-sensor, to multi-agent shared perception, with benchmark evidence that fusion stops detection collapse under corruption.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

A safety-critical survey that organizes BEV perception into single-modality, multimodal, and collaborative stages and consolidates robustness evidence that multimodal fusion degrades far less than single-modality perception under sensor corruption.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection Useful SafeBEV taxonomy and broad coverage, but Table 6 and the BEVFormer 'multimodal' claim are internally inconsistent; needs revision before it can serve as a trusted reference. the 3 major comments →

arxiv 2508.07560 v1 pith:HHMFGH2J submitted 2025-08-11 cs.RO cs.CV

Progressive Bird's Eye View Perception for Safety-Critical Autonomous Driving: A Comprehensive Survey

Yan Gong , Naibang Wang , Jianli Lu , Xinyu Zhang , Yongsheng Gao , Jie Zhao , Zifan Huang , Haozhi Bai
show 8 more authors
Nanxin Zeng Nayu Su Lei Yang Ziying Song Xiaoxi Hu Xinmin Jiang Xiaojuan Zhang Susanto Rahardja
This is my paper
classification cs.RO cs.CV
keywords bird's-eye-view perceptionautonomous drivingmultimodal sensor fusioncollaborative perceptionsafety-critical systemsrobustness benchmarkscorruption robustnessV2X communication
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The reading

This survey set out to organize the bird's-eye-view (BEV) perception field for autonomous driving around one question: what makes BEV perception safe enough for real roads? It claims to be the first systematic review of BEV perception from a safety-critical perspective, and it structures its answer as three progressive stages—single-modality vehicle-side perception, multimodal vehicle-side fusion, and multi-agent collaborative perception—covering well over one hundred published methods and dozens of public datasets. The quantitative centerpiece is a robustness comparison: on the nuScenes-C and BEV-Robust corruption benchmarks, single-modality models such as BEVDet and BEVDepth are reported to collapse (BEVDet's mAP falls to 3.3% under noise), while multimodal fusion models such as BEVFusion and TransFusion keep mAP and NDS above 60% across corruptions. If the survey is right, the field gains a safety-oriented reference taxonomy, a dataset map graded by safety relevance, and evidence that sensor redundancy—not any single sensor—is the main defense against perception collapse in adverse conditions.

Core claim

The paper's central claim: BEV perception has evolved through three stages of safety, and robustness increases at each step. In SafeBEV 1.0 a single sensor (camera or LiDAR) builds the top-down BEV map; in SafeBEV 2.0 cameras, LiDAR, and radar are fused in that shared BEV coordinate system; in SafeBEV 3.0 BEV features are exchanged and fused across vehicles and roadside infrastructure (V2V, V2I, V2X, I2I). The survey's evidence for the stage-wise robustness gain is a benchmark comparison following the nuScenes-C and BEV-Robust corruption protocols. It reports that single-modality detectors collapse—BEVDet reaches only 3.3% mAP under noise and 11.6% under blur, with NDS below 11%—while BEVFus

What carries the argument

Two pieces of machinery carry the argument. The first is the SafeBEV taxonomy: the bird's-eye view representation—a top-down grid in which multi-camera, LiDAR, and radar inputs are projected into one shared spatial coordinate system—plus the three-stage progression that organizes every method reviewed (single-modality, multimodal fusion, multi-agent collaboration). The taxonomy gives the survey its structure and its claim that robustness increases across stages. The second is the robustness benchmarking protocol: the nuScenes-C and BEV-Robust standardized corruption sets (camera failure, quantization, motion blur, noise, digital artifacts, weather) together with the metrics mAP, NDS, mCE, an

Load-bearing premise

The load-bearing premise is that Table 6 and the Section 5.4.2 prose faithfully transcribe the published nuScenes-C and BEV-Robust benchmark numbers; the table's BEVDet row merges cells, and its BEVFormer mRR values sit at 52.1–67.8 while the prose says they stay above 96%, so if the transcription is wrong the survey's central robustness comparison is unsupported.

What would settle it

Open the original nuScenes-C and BEV-Robust result tables (references [294] and [295]) and check three numbers: BEVDet's BEV-Robust blur mAP/NDS (printed here as '11.6/20.6521.22/30.523.8/33.8'), BEVDet's noise mAP (3.3%), and BEVFormer's nuScenes-C mRR values (printed here as 52.1–67.8 while the prose says above 96%). If the original tables match what Table 6 prints, the central claim stands on its data; if they do not, the survey's headline robustness result has not been reliably reported.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • If the robustness comparison is right, the practical route to avoiding perception collapse in bad weather, noise, or blur is to equip vehicles with complementary sensors and fuse them in BEV space—a direct argument for camera–LiDAR–radar suites over single-sensor designs.
  • If the taxonomy is right, any new BEV method, dataset, or open problem can be placed on a three-stage progress map that makes the remaining safety gaps visible.
  • The safety-relevance grading of datasets tells practitioners which benchmark exercises which failure mode (extreme weather, occlusion, sensor noise, accident scenarios), easing dataset choice beyond raw task fit.
  • The open challenges named in the survey delimit the near-term research agenda: open-set recognition, semi-supervised and synthetic data for multi-agent learning, corruption-aware training, uncertainty modeling, and latency-aware collaborative fusion.
  • The accompanying open-source repository turns the survey's organization into a working resource for reproducing method and dataset comparisons.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The stage-wise framing suggests a monotonic robustness hierarchy (collaboration > multimodal fusion > single modality), but the paper only benchmarks vehicle-side models; running the same nuScenes-C/BEV-Robust corruptions on V2X collaborative models—with added communication latency and packet loss—would supply the missing SafeBEV 3.0 robustness evidence.
  • Table 6's BEVDet row merges several cells and its BEVFormer mRR values contradict the Section 5.4.2 claim, so the headline robustness figures deserve re-checking against the original [294] and [295] tables before being quoted; the qualitative claim likely survives, but specific percentages may not.
  • The sensor-failure formula in Section 6.3, 1-(1-p)^n, points to a design tension the survey leaves implicit: more sensors mean a higher probability that at least one input is corrupted, so the value of redundancy should be benchmarked against single-modality dropout, not just additive corruption.
  • A testable extension of the dataset analysis would map each of the six safety scenario families (Fig. 8) onto a concrete evaluation protocol, so that 'dataset supports extreme weather' becomes a measurable claim rather than a qualitative one.
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

3 major / 4 minor

Summary. This survey reviews bird's-eye-view (BEV) perception for autonomous driving from a safety- and robustness-oriented perspective. It organizes the literature into three progressive stages — SafeBEV 1.0 (single-modality vehicle-side), SafeBEV 2.0 (multimodal vehicle-side), and SafeBEV 3.0 (multi-agent collaborative perception) — and covers representative methods, public datasets, robustness benchmarks, open challenges, and future directions. The authors also provide an open-source repository. The quantitative centerpiece is Section 5.4, where results from nuScenes-C and BEV-Robust are used to argue that single-modality perception collapses under corruption while multimodal fusion maintains high mAP/NDS.

Significance. If the survey's methodological classifications and quantitative comparisons are reliable, the paper would be a useful safety-oriented reference for the BEV perception community. Its strengths are the breadth of coverage (roughly 200 methods and 50 datasets), the explicit three-stage taxonomy, the recent references, and the accompanying open-source repository. However, the central quantitative evidence in Table 6 and Section 5.4.2 contains internal contradictions and corrupted table entries. Because the headline conclusion about the robustness advantage of multimodal fusion depends on these numbers, the manuscript is not yet trustworthy as a secondary source for benchmark results. The organizational and qualitative content is valuable; the quantitative centerpiece needs correction and verification.

major comments (3)
  1. [Section 5.4.2 and Table 6] There is a direct internal contradiction. The text states: 'BEVFormer also performs robustly under nuScenes-C corruptions, maintaining mCE below 102% and mRR above 96%.' However, Table 6 lists BEVFormer's nuScenes-C mRR values as 59.6, 67.8, 52.1, and 61.0 — all far below 96%. The same table does show mCE values below 102%, so the prose appears to conflate mCE and mRR. This is load-bearing because the paragraph uses these numbers to support the conclusion that multimodal/temporal fusion methods are resilient. The authors must correct the prose, the table, or both and verify every entry against the original nuScenes-C [294] and BEV-Robust [295] sources.
  2. [Table 6, BEVDet row] The BEVDet row contains a corrupted merged cell: '3.3/10.5 11.6/20.6521.22/30.523.8/33.8 29.2/37.2'. The blur, digital, and weather entries are concatenated without separators, making the row unreadable and the reported values ambiguous. This is not merely a typo: it directly affects the paper's central quantitative comparison and raises concerns about the reliability of the entire table's transcription. Please regenerate the table and add a machine-readable or clearly formatted version.
  3. [Section 5.3 and Figure 8] The radar charts in Figure 8 purport to show dataset support for six safety-critical capabilities, and Section 5.3 uses them to draw conclusions such as 'most datasets still offer limited support for high-risk scenarios.' However, no scoring rubric, annotation protocol, or source is provided for these values. Without a defined scale or a transparent methodology, the quantitative appearance of these charts is not reproducible. Either replace them with documented categorical assessments or provide the scoring criteria and the per-dataset evidence.
minor comments (4)
  1. [Section 2.1 vs Table 6] BEVFormer is described in Section 2 as a camera-based single-modality method, yet Table 6 categorizes it under SafeBEV 2.0 (multimodal/temporal fusion). Since the SafeBEV taxonomy is one of the paper's main contributions, this inconsistency should be clarified. The original BEVFormer is camera-only; if the authors intend to treat temporal fusion as a distinct 'modality' dimension, that criterion should be stated explicitly.
  2. [Section 5.4.2] The sentence 'On nuScenes-C, models like BEVDet and BEVerse report mCE values exceeding 120%, with mRR declines of more than 50%' is not directly supported by Table 6 as printed. BEVDet's mCE reaches 121.8 in one setting and BEVerse reaches 121.8 in another, but the mRR values in the table are 44.8–66.7. If 'decline' is computed relative to clean performance, the definition should be given and the numbers recomputed.
  3. [Tables 4 and 5] Several table cells are ambiguous due to compressed formatting. For example, Table 4's 'Frames/Annotation' columns show entries like '8K 15 - 15' for CamVid and '133K - 8 4' for Lane Det, with no clear indication of which number corresponds to which column. Table 5 has similar issues in the Size/Agents columns. Please add clear column headers and units.
  4. [General] There are small typographical errors: the author email 'su11031915@163,com' uses a comma instead of a dot, and Section 5.2.1 spells 'RCooper' as 'RCooperr'. These are minor but should be fixed.

Circularity Check

0 steps flagged

No significant circularity: the survey's central robustness conclusion is anchored in external benchmarks; the only self-citation is non-load-bearing.

full rationale

This is a survey, not a derivation chain. The only quantitative inference in the paper—that single-modality BEV models collapse under corruption while multimodal fusion retains mAP/NDS above 60%—is explicitly drawn from two external benchmarks, nuScenes-C [294] and BEV-Robust [295] (Section 5.4, Table 6). The table's corrupted cells (e.g., BEVDet '11.6/20.6521.22/30.523.8/33.8') and the contradiction between the prose claim 'mRR above 96%' for BEVFormer and the listed 52.1–67.8 mRR values are transcription/reporting errors; they undermine factual reliability but do not make the conclusion an input to itself. The SafeBEV 1.0/2.0/3.0 framework is an organizing taxonomy imposed by the authors, not a result derived from a fitted parameter; categorizing methods into self-defined stages is definitional only in the benign sense of any survey structure. The only self-citation is ref [8] (Wang et al. 2025, overlapping authors), cited in Section 5's opening alongside two other references to support the general statement that high-quality datasets are essential for perception evaluation; the dataset tables are compiled from primary dataset papers, and the central safety-robustness conclusion does not depend on ref [8]. No load-bearing self-citation, no imported uniqueness theorem, no ansatz smuggled in via citation, and no fitted input renamed as prediction. The benchmark-number corruption flagged by the reader is a correctness risk, not circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 3 axioms · 1 invented entities

The survey does not fit parameters or postulate physical entities; its free-parameter count is zero. It does make load-bearing assumptions: (1) the three-stage taxonomy is the right organizing frame for safety (Section 1, Fig. 1); (2) the consolidated benchmark table (Table 6) accurately reflects nuScenes-C and BEV-Robust; (3) the first comprehensive safety-critical survey claim is valid relative to the cited survey landscape; and (4) Section 5's dataset catalog, drawn in large part from the authors' own prior survey (ref [8]), is accurate. The only invented entity is the SafeBEV 1.0/2.0/3.0 framework, a naming and organization device with no independent evidence. These are typical axioms for a review; they transfer the survey's credibility burden to the accuracy of its restatements.

axioms (3)
  • domain assumption The three-stage progression (single-modality, multimodal, multi-agent collaborative) is the right organizing axis for safety and robustness
    Stated in Section 1 and Fig. 1 as the field's evolution toward safety; nowhere derived. A survey's value depends on this frame being useful rather than merely imposed on the literature.
  • domain assumption Table 6 faithfully reproduces the nuScenes-C and BEV-Robust benchmark numbers
    Section 5.4.2 draws its central conclusions (fusion maintains accuracy above 60%, single-modality collapses) from this table. The table contains a merged-entry formatting corruption and a prose/table contradiction on BEVFormer mRR, so the table cannot be trusted as printed without cross-checking the original benchmark sources [294], [295].
  • domain assumption The first comprehensive safety-critical survey claim is valid relative to the cited survey landscape
    Section 1 identifies gaps in Ma et al. [96], Li et al. [23], Zhao et al. [24], and Singh et al. [14]. The first claim is only as strong as this comparison, and the authors' own prior dataset survey [8] overlaps substantially with Section 5.
invented entities (1)
  • SafeBEV 1.0/2.0/3.0 progressive-stage taxonomy no independent evidence
    purpose: Organizes the survey and the historical timeline of BEV perception into three maturity stages (single-modality, multimodal, multi-agent collaboration)
    A framing and branding device introduced by this paper (Section 1, Fig. 1, Fig. 2). It carries no falsifiable predictions and cannot be verified independently; its value is purely organizational.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Progressive Bird's Eye View Perception for Safety-Critical Autonomous Driving: A Comprehensive Survey." pith.science (2026). https://pith.science/paper/HHMFGH2J

@misc{pith2026250807560,
  author       = {Pith},
  title        = {Pith review of: Progressive Bird's Eye View Perception for Safety-Critical Autonomous Driving: A Comprehensive Survey},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HHMFGH2J}},
  note         = {Machine review of arXiv:2508.07560}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Bird's-Eye-View (BEV) perception has become a foundational paradigm in autonomous driving, enabling unified spatial representations that support robust multi-sensor fusion and multi-agent collaboration. As autonomous vehicles transition from controlled environments to real-world deployment, ensuring the safety and reliability of BEV perception in complex scenarios - such as occlusions, adverse weather, and dynamic traffic - remains a critical challenge. This survey provides the first comprehensive review of BEV perception from a safety-critical perspective, systematically analyzing state-of-the-art frameworks and implementation strategies across three progressive stages: single-modality vehicle-side, multimodal vehicle-side, and multi-agent collaborative perception. Furthermore, we examine public datasets encompassing vehicle-side, roadside, and collaborative settings, evaluating their relevance to safety and robustness. We also identify key open-world challenges - including open-set recognition, large-scale unlabeled data, sensor degradation, and inter-agent communication latency - and outline future research directions, such as integration with end-to-end autonomous driving systems, embodied intelligence, and large language models.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

300 extracted references · 35 canonical work pages · 19 internal anchors

  1. [1]

    IEEE Transactions on Circuits and Systems for Video Technology34(7), 5753–5764 (2024)

    Wang, J., Li, F., An, Y., Zhang, X., Sun, H.: Toward robust lidar-camera fusion in bev space via mutual deformable attention and temporal aggregation. IEEE Transactions on Circuits and Systems for Video Technology34(7), 5753–5764 (2024)

    2024

  2. [2]

    IEEE Transactions on Circuits and Systems for Video Technology34(4), 2619–2632 (2023)

    Song, Z., Jia, C., Yang, L., Wei, H., Liu, L.: Graphalign++: An accurate feature align- ment by graph matching for multi-modal 3d object detection. IEEE Transactions on Circuits and Systems for Video Technology34(4), 2619–2632 (2023)

    2023

  3. [3]

    IEEE Transactions on Intelligent Vehicles (2024)

    Gong, Y., Jiang, X., Wang, L., Xu, L., Lu, J., Liu, H., Lin, L., Zhang, X.: Tclanenet: Task- conditioned lane detection network driven by vibration information. IEEE Transactions on Intelligent Vehicles (2024)

    2024

  4. [4]

    IEEE Transactions on Intelligent Vehicles 9(2), 4025–4039 (2023)

    Zhang, X., Gong, Y., Lu, J., Li, Z., Li, S., Wang, S., Liu, W., Wang, L., Li, J.: Oblique convolution: A novel convolution idea for redefining lane detection. IEEE Transactions on Intelligent Vehicles 9(2), 4025–4039 (2023)

    2023

  5. [5]

    Engineering Applications of Artificial Intelligence 129, 107575 (2024)

    Liu, W., Gong, Y., Zhang, G., Lu, J., Zhou, Y., Liao, J.: Glmdrivenet: Global–local multimodal fusion driving behavior classification network. Engineering Applications of Artificial Intelligence 129, 107575 (2024)

    2024

  6. [6]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Sun, P., Kretzschmar, H., Dotiwalla, X., Chouard, A., Patnaik, V., Tsui, P., Guo, J., Zhou, Y., Chai, Y., Caine, B.,et al.: Scalability in perception for autonomous driving: Waymo open dataset. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 2446–2454 (2020)

    2020

  7. [7]

    IEEE Transactions on Neural Networks and Learning Systems 32(8), 3412–3432 (2020)

    Li, Y., Ma, L., Zhong, Z., Liu, F., Chapman, M.A., Cao, D., Li, J.: Deep learning for lidar point clouds in autonomous driving: A review. IEEE Transactions on Neural Networks and Learning Systems 32(8), 3412–3432 (2020)

    2020

  8. [8]

    arXiv preprint arXiv:2504.12696 (2025)

    Wang, N., Shang, D., Gong, Y., Hu, X., Song, Z., Yang, L., Huang, Y., Wang, X., Lu, J.: Collaborative perception datasets for autonomous driving: A review. arXiv preprint arXiv:2504.12696 (2025)

    Pith/arXiv arXiv 2025

  9. [9]

    In: 2020 IEEE 16th International Conference 36 on Intelligent Computer Communication and Processing (ICCP), pp

    Velasco-Hernandez, G., Barry, J., Walsh, J., et al.: Autonomous driving architectures, perception and data fusion: A review. In: 2020 IEEE 16th International Conference 36 on Intelligent Computer Communication and Processing (ICCP), pp. 315–321 (2020). IEEE

    2020

  10. [10]

    IEEE Transactions on Intelligent Vehicles8(6), 3605–3619 (2023)

    Zhang, X., Gong, Y., Lu, J., Wu, J., Li, Z., Jin, D., Li, J.: Multi-modal fusion technology based on vehicle information: A survey. IEEE Transactions on Intelligent Vehicles8(6), 3605–3619 (2023)

    2023

  11. [11]

    RoboFusion: Towards Robust Multi-Modal 3D Object Detection via SAM

    Song, Z., Zhang, G., Liu, L., Yang, L., Xu, S., Jia, C., Jia, F., Wang, L.: Robo- fusion: Towards robust multi-modal 3d obiect detection via sam. arXiv preprint arXiv:2401.03907 (2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  12. [12]

    IEEE Transactions on Circuits and Systems for Video Technology (2024)

    Lu, Z., Cao, B., Hu, Q.: Lidar-camera continuous fusion in voxelized grid for semantic scene completion. IEEE Transactions on Circuits and Systems for Video Technology (2024)

    2024

  13. [13]

    2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC), 5292–5299 (2023) https://doi.org/10.1109/ITSC57777.2023.10421963

    Xiang, C., Xie, X., Feng, C., Bai, Z., Niu, Z., Yang, M.: V2I-BEVF: Multi-modal Fusion Based on BEV Representation for Vehicle-Infrastructure Perception. 2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC), 5292–5299 (2023) https://doi.org/10.1109/ITSC57777.2023.10421963

    arXiv 2023

  14. [14]

    In: 2023 IEEE Intelligent Vehicles Symposium (IV), pp

    Singh, A.: Vision-radar fusion for robotics bev detections: A survey. In: 2023 IEEE Intelligent Vehicles Symposium (IV), pp. 1–7 (2023). IEEE

    2023

  15. [15]

    IEEE Transactions on Intelligent Transportation Systems 24(12), 14121–14132 (2023)

    Li, R., Dong, Y.: Robust lane detection through self pre-training with masked sequen- tial autoencoders and fine-tuning with customized polyloss. IEEE Transactions on Intelligent Transportation Systems 24(12), 14121–14132 (2023)

    2023

  16. [16]

    IEEE Transactions on Intelligent Transportation Systems (2024)

    Gong, Y., Zhang, X., Lu, J., Jiang, X., Wang, Z., Liu, H., Li, Z., Wang, L., Yang, Q., Wu, X.: Steering angle-guided multimodal fusion lane detection for autonomous driving. IEEE Transactions on Intelligent Transportation Systems (2024)

    2024

  17. [17]

    In: 2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC), pp

    Zhao, Z., Li, Y., Chen, Y., Zhang, X., Tian, R.: A spatial alignment framework using geolocation cues for roadside multi-view multi-sensor fusion. In: 2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC), pp. 3633–3640 (2023). IEEE

    2023

  18. [18]

    Advances in Neural Information Processing Systems 34, 29541–29552 (2021)

    Li, Y., Ren, S., Wu, P., Chen, S., Feng, C., Zhang, W.: Learning distilled collaboration graph for multi-agent perception. Advances in Neural Information Processing Systems 34, 29541–29552 (2021)

    2021

  19. [19]

    In: Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part II 16, pp

    Wang, T.-H., Manivasagam, S., Liang, M., Yang, B., Zeng, W., Urtasun, R.: V2vnet: Vehicle-to-vehicle communication for joint perception and prediction. In: Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part II 16, pp. 605–621 (2020). Springer

    2020

  20. [20]

    Journal of Dynamic 37 Systems, Measurement, and Control 143(7), 071002 (2021)

    Viana, I.B., Kanchwala, H., Ahiska, K., Aouf, N.: A comparison of trajectory plan- ning and control frameworks for cooperative autonomous driving. Journal of Dynamic 37 Systems, Measurement, and Control 143(7), 071002 (2021)

    2021

  21. [21]

    In: European Conference on Computer Vision, pp

    Song, Z., Yang, L., Xu, S., Liu, L., Xu, D., Jia, C., Jia, F., Wang, L.: Graphbev: Towards robust bev feature alignment for multi-modal 3d object detection. In: European Conference on Computer Vision, pp. 347–366 (2025). Springer

    2025

  22. [22]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Man, Y., Gui, L.-Y., Wang, Y.-X.: Bev-guided multi-modality fusion for driving per- ception. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 21960–21969 (2023)

    2023

  23. [23]

    IEEE Transactions on Pattern Analysis and Machine Intelligence 46(4), 2151–2170 (2023)

    Li, H., Sima, C., Dai, J., Wang, W., Lu, L., Wang, H., Zeng, J., Li, Z., Yang, J., Deng, H., et al.: Delving into the devils of bird’s-eye-view perception: A review, evaluation and recipe. IEEE Transactions on Pattern Analysis and Machine Intelligence 46(4), 2151–2170 (2023)

    2023

  24. [24]

    Expert Systems with Applications 258, 125103 (2024)

    Zhao, J., Shi, J., Zhuo, L.: Bev perception for autonomous driving: State of the art and future perspectives. Expert Systems with Applications 258, 125103 (2024)

    2024

  25. [25]

    IEEE Transactions on Intelligent Transportation Systems 22(3), 1341–1360 (2020)

    Feng, D., Haase-Sch¨ utz, C., Rosenbaum, L., Hertlein, H., Glaeser, C., Timm, F., Wies- beck, W., Dietmayer, K.: Deep multi-modal object detection and semantic segmentation for autonomous driving: Datasets, methods, and challenges. IEEE Transactions on Intelligent Transportation Systems 22(3), 1341–1360 (2020)

    2020

  26. [26]

    arXiv preprint arXiv:2207.02202 (2022)

    Xu, R., Tu, Z., Xiang, H., Shao, W., Zhou, B., Ma, J.: Cobevt: Cooperative bird’s eye view semantic segmentation with sparse transformers. arXiv preprint arXiv:2207.02202 (2022)

    Pith/arXiv arXiv 2022

  27. [27]

    In: 2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC), pp

    Xiang, C., Xie, X., Feng, C., Bai, Z., Niu, Z., Yang, M.: V2i-bevf: Multi-modal fusion based on bev representation for vehicle-infrastructure perception. In: 2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC), pp. 5292–5299 (2023). IEEE

    2023

  28. [28]

    arXiv preprint arXiv:2310.06008 (2023)

    Qiao, D., Zulkernine, F.: Cobevfusion: Cooperative perception with lidar-camera bird’s- eye view fusion. arXiv preprint arXiv:2310.06008 (2023)

    Pith/arXiv arXiv 2023

  29. [29]

    Transportation Research Record 2679(5), 319–334 (2025)

    Dong, Y., Zhang, L., Farah, H., Zgonnikov, A., Arem, B.: Data-driven semi-supervised machine learning with safety indicators for abnormal driving behavior detection. Transportation Research Record 2679(5), 319–334 (2025)

    2025

  30. [30]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Yang, L., Yu, K., Tang, T., Li, J., Yuan, K., Wang, L., Zhang, X., Chen, P.: Bevheight: A robust framework for vision-based roadside 3d object detection. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 21611–21620 (2023)

    2023

  31. [31]

    In: 2023 IEEE International Conference on Robotics and Automation (ICRA), pp

    Zou, J., Zhu, Z., Huang, J., Yang, T., Huang, G., Wang, X.: Hft: Lifting perspective representations via hybrid feature transformation for bev perception. In: 2023 IEEE International Conference on Robotics and Automation (ICRA), pp. 7046–7053 (2023). 38 IEEE

    2023

  32. [32]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Wang, Y., Chao, W.-L., Garg, D., Hariharan, B., Campbell, M., Weinberger, K.Q.: Pseudo-lidar from visual depth estimation: Bridging the gap in 3d object detection for autonomous driving. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 8445–8453 (2019)

    2019

  33. [33]

    IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

    Li, Z., Wang, W., Li, H., Xie, E., Sima, C., Lu, T., Yu, Q., Dai, J.: Bevformer: learn- ing bird’s-eye-view representation from lidar-camera via spatiotemporal transformers. IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

    2024

  34. [34]

    In: 2024 International Conference on 3D Vision (3DV), pp

    Sautier, C., Puy, G., Boulch, A., Marlet, R., Lepetit, V.: Bevcontrast: Self-supervision in bev space for automotive lidar point clouds. In: 2024 International Conference on 3D Vision (3DV), pp. 559–568 (2024). IEEE

    2024

  35. [35]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Shi, S., Wang, X., Li, H.: Pointrcnn: 3d object proposal generation and detection from point cloud. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 770–779 (2019)

    2019

  36. [36]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Lang, A.H., Vora, S., Caesar, H., Zhou, L., Yang, J., Beijbom, O.: Pointpillars: Fast encoders for object detection from point clouds. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 12697–12705 (2019)

    2019

  37. [37]

    arXiv preprint arXiv:2411.08402 (2024)

    Huang, X., Wang, J., Xia, Q., Chen, S., Yang, B., Wang, C., Wen, C.: V2x-r: Cooperative lidar-4d radar fusion for 3d object detection with denoising diffusion. arXiv preprint arXiv:2411.08402 (2024)

    Pith/arXiv arXiv 2024

  38. [38]

    IEEE Robotics and Automation Letters 5(2), 1263–1270 (2020)

    Palffy, A., Dong, J., Kooij, J.F., Gavrila, D.M.: Cnn based road user detection using the 3d radar cube. IEEE Robotics and Automation Letters 5(2), 1263–1270 (2020)

    2020

  39. [39]

    In: 2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC), pp

    Zheng, L., Ma, Z., Zhu, X., Tan, B., Li, S., Long, K., Sun, W., Chen, S., Zhang, L., Wan, M., et al.: Tj4dradset: A 4d radar dataset for autonomous driving. In: 2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC), pp. 493–498 (2022). IEEE

    2022

  40. [40]

    CALICO: Self-Supervised Camera-LiDAR Contrastive Pre-training for BEV Perception

    Sun, J., Zheng, H., Zhang, Q., Prakash, A., Mao, Z.M., Xiao, C.: Calico: Self- supervised camera-lidar contrastive pre-training for bev perception. arXiv preprint arXiv:2306.00349 (2023)

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  41. [41]

    arXiv preprint arXiv:2203.17054 (2022)

    Huang, J., Huang, G.: Bevdet4d: Exploit temporal cues in multi-camera 3d object detection. arXiv preprint arXiv:2203.17054 (2022)

    Pith/arXiv arXiv 2022

  42. [42]

    Authorea Preprints

    Li, H., Zhao, Y., Zhong, J., Wang, B., Sun, C., Sun, F.: Delving into the secrets of bev 3d object detection in autonomous driving: A comprehensive survey. Authorea Preprints

  43. [43]

    39 IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

    Li, Y., Huang, B., Chen, Z., Cui, Y., Liang, F., Shen, M., Liu, F., Xie, E., Sheng, L., Ouyang, W., et al.: Fast-bev: A fast and strong bird’s-eye view perception baseline. 39 IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

    2024

  44. [44]

    IEEE Transactions on Intelligent Vehicles8(11), 4498–4514 (2023)

    Chang, C., Zhang, J., Zhang, K., Zhong, W., Peng, X., Li, S., Li, L.: Bev-v2x: Cooper- ative birds-eye-view fusion and grid occupancy prediction via v2x-based data sharing. IEEE Transactions on Intelligent Vehicles8(11), 4498–4514 (2023)

    2023

  45. [45]

    IEEE Transactions on Intelligent Vehicles (2024)

    Wang, P., Zhu, M., Zheng, X., Lu, H., Zhong, H., Chen, X., Shen, S., Wang, X., Wang, Y., Wang, F.-Y.: Bevgpt: Generative pre-trained foundation model for autonomous driving prediction, decision-making, and planning. IEEE Transactions on Intelligent Vehicles (2024)

    2024

  46. [46]

    Hierarchical End-to-End Autonomous Driving: Integrating BEV Perception with Deep Reinforcement Learning

    Lu, S., He, L., Li, S.E., Luo, Y., Wang, J., Li, K.: Hierarchical end-to-end autonomous driving: Integrating bev perception with deep reinforcement learning. arXiv preprint arXiv:2409.17659 (2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  47. [47]

    IEEE Robotics and Automation Letters (2024)

    Jiang, J., Yang, Y., Deng, Y., Ma, C., Zhang, J.: Bevnav: Robot autonomous naviga- tion via spatial-temporal contrastive learning in bird’s-eye view. IEEE Robotics and Automation Letters (2024)

    2024

  48. [48]

    arXiv preprint arXiv:2501.11260 (2025)

    Feng, T., Wang, W., Yang, Y.: A survey of world models for autonomous driving. arXiv preprint arXiv:2501.11260 (2025)

    Pith/arXiv arXiv 2025

  49. [49]

    arXiv preprint arXiv:2112.11790 (2021)

    Huang, J., Huang, G., Zhu, Z., Ye, Y., Du, D.: Bevdet: High-performance multi-camera 3d object detection in bird-eye-view. arXiv preprint arXiv:2112.11790 (2021)

    Pith/arXiv arXiv 2021

  50. [50]

    In: 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp

    Sun, R., Fu, M., Liang, H., Zhu, C., Dong, Z., Yang, Y.: Robust multi-camera bev perception: An image-perceptive approach to counter imprecise camera calibration. In: 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 5002–5008 (2024). IEEE

    2024

  51. [51]

    In: European Conference on Computer Vision, pp

    Liu, C., Zhu, M., Ma, C.: H-v2x: A large scale highway dataset for bev perception. In: European Conference on Computer Vision, pp. 139–157 (2024). Springer

    2024

  52. [52]

    Image and vision computing 16(8), 585–590 (1998)

    Bertozz, M., Broggi, A., Fascioli, A.: Stereo inverse perspective mapping: theory and applications. Image and vision computing 16(8), 585–590 (1998)

    1998

  53. [53]

    Applied Intelligence 53(19), 22117–22131 (2023)

    Gong, Y., Wang, L., Xu, L.: A feature aggregation network for multispectral pedestrian detection. Applied Intelligence 53(19), 22117–22131 (2023)

    2023

  54. [54]

    Automotive Innovation, 1–17 (2025)

    Gong, Y., Zhang, X., Liu, H., Jiang, X., Li, Z., Gao, X., Lin, L., Jin, D., Li, J., Liu, H.: Skipcrossnets: Adaptive skip-cross fusion for road detection. Automotive Innovation, 1–17 (2025)

    2025

  55. [55]

    In: Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XIV 16, pp

    Philion, J., Fidler, S.: Lift, splat, shoot: Encoding images from arbitrary camera rigs by implicitly unprojecting to 3d. In: Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XIV 16, pp. 194–210 (2020). Springer 40

    2020

  56. [56]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp

    Hu, A., Murez, Z., Mohan, N., Dudas, S., Hawke, J., Badrinarayanan, V., Cipolla, R., Kendall, A.: Fiery: Future instance prediction in bird’s-eye view from surround monocular cameras. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 15273–15282 (2021)

    2021

  57. [57]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Roddick, T., Cipolla, R.: Predicting semantic map representations from images using pyramid occupancy networks. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 11138–11147 (2020)

    2020

  58. [58]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Yang, W., Li, Q., Liu, W., Yu, Y., Ma, Y., He, S., Pan, J.: Projecting your view attentively: Monocular road scene layout estimation via cross-view transformation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 15536– 15545 (2021)

    2021

  59. [59]

    In: 2022 International Conference on Robotics and Automation (ICRA), pp

    Saha, A., Mendez, O., Russell, C., Bowden, R.: Translating images into maps. In: 2022 International Conference on Robotics and Automation (ICRA), pp. 9200–9206 (2022). IEEE

    2022

  60. [60]

    In: Proceedings of the 30th ACM International Conference on Multimedia, pp

    Chen, Z., Li, Z., Zhang, S., Fang, L., Jiang, Q., Zhao, F.: Graph-detr3d: rethinking overlapping regions for multi-view 3d object detection. In: Proceedings of the 30th ACM International Conference on Multimedia, pp. 5999–6008 (2022)

    2022

  61. [61]

    In: European Conference on Computer Vision, pp

    Liu, Y., Wang, T., Zhang, X., Sun, J.: Petr: Position embedding transformation for multi- view 3d object detection. In: European Conference on Computer Vision, pp. 531–548 (2022). Springer

    2022

  62. [62]

    In: Proceedings of the AAAI Conference on Artificial Intelligence, vol

    Li, Y., Ge, Z., Yu, G., Yang, J., Wang, Z., Shi, Y., Sun, J., Li, Z.: Bevdepth: Acquisition of reliable depth for multi-view 3d object detection. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, pp. 1477–1485 (2023)

    2023

  63. [63]

    arXiv preprint arXiv:2302.02367 (2023)

    Zhou, S., Tian, Z., Chu, X., Zhang, X., Zhang, B., Lu, X., Feng, C., Jie, Z., Chiang, P.Y., Ma, L.: Fastpillars: a deployment-friendly pillar-based 3d detector. arXiv preprint arXiv:2302.02367 (2023)

    Pith/arXiv arXiv 2023

  64. [64]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Zhang, G., Chen, J., Gao, G., Li, J., Liu, S., Hu, X.: Safdnet: A simple and effective net- work for fully sparse 3d object detection. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 14477–14486 (2024)

    2024

  65. [65]

    In: Proceedings of the AAAI Conference on Artificial Intelligence, vol

    Qiu, S., Li, X., Xue, X., Pu, J.: Pc-bev: An efficient polar-cartesian bev fusion framework for lidar semantic segmentation. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 39, pp. 6612–6620 (2025)

    2025

  66. [66]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Shi, S., Guo, C., Jiang, L., Wang, Z., Shi, J., Wang, X., Li, H.: Pv-rcnn: Point-voxel fea- ture set abstraction for 3d object detection. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 10529–10538 (2020)

    2020

  67. [67]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp

    Qi, C.R., Su, H., Mo, K., Guibas, L.J.: Pointnet: Deep learning on point sets for 3d 41 classification and segmentation. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 652–660 (2017)

    2017

  68. [68]

    Engineering Applications of Artificial Intelligence 133, 108550 (2024)

    Thakur, A., Mishra, S.K.: An in-depth evaluation of deep learning-enabled adaptive approaches for detecting obstacles using sensor-fused data in autonomous vehicles. Engineering Applications of Artificial Intelligence 133, 108550 (2024)

    2024

  69. [69]

    In: Proceedings of the AAAI Conference on Artificial Intelligence, vol

    Kim, Y., Kim, S., Choi, J.W., Kum, D.: Craft: Camera-radar 3d object detection with spatio-contextual fusion transformer. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, pp. 1160–1168 (2023)

    2023

  70. [70]

    IEEE Transactions on Intelligent Vehicles 9(1), 1524–1536 (2023)

    Yu, Z., Wan, W., Ren, M., Zheng, X., Fang, Z.: Sparsefusion3d: Sparse sensor fusion for 3d object detection by radar and camera in environmental perception. IEEE Transactions on Intelligent Vehicles 9(1), 1524–1536 (2023)

    2023

  71. [71]

    In: 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp

    Schramm, J., V¨odisch, N., Petek, K., Kiran, B.R., Yogamani, S., Burgard, W., Valada, A.: Bevcar: Camera-radar fusion for bev map and object segmentation. In: 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 1435–1442 (2024). IEEE

    2024

  72. [72]

    In: 2023 IEEE International Conference on Robotics and Automation (ICRA), pp

    Liu, Z., Tang, H., Amini, A., Yang, X., Mao, H., Rus, D.L., Han, S.: Bevfusion: Multi- task multi-sensor fusion with unified bird’s-eye view representation. In: 2023 IEEE International Conference on Robotics and Automation (ICRA), pp. 2774–2781 (2023). IEEE

    2023

  73. [73]

    Advances in Neural Information Processing Systems 35, 10421–10434 (2022)

    Liang, T., Xie, H., Yu, K., Xia, Z., Lin, Z., Wang, Y., Tang, T., Wang, B., Tang, Z.: Bevfusion: A simple and robust lidar-camera fusion framework. Advances in Neural Information Processing Systems 35, 10421–10434 (2022)

    2022

  74. [74]

    arXiv preprint arXiv:2309.05257 (2023)

    Hu, C., Zheng, H., Li, K., Xu, J., Mao, W., Luo, M., Wang, L., Chen, M., Peng, Q., Liu, K., et al.: Fusionformer: A multi-sensory fusion in bird’s-eye-view and temporal consistent transformer for 3d object detection. arXiv preprint arXiv:2309.05257 (2023)

    Pith/arXiv arXiv 2023

  75. [75]

    In: European Conference on Computer Vision, pp

    Yang, B., Guo, R., Liang, M., Casas, S., Urtasun, R.: Radarnet: Exploiting radar for robust perception of dynamic objects. In: European Conference on Computer Vision, pp. 496–512 (2020). Springer

    2020

  76. [76]

    In: 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp

    Wang, L., Zhang, X., Xv, B., Zhang, J., Fu, R., Wang, X., Zhu, L., Ren, H., Lu, P., Li, J., et al.: Interfusion: Interaction-based 4d radar and lidar fusion for 3d object detection. In: 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 12247–12253 (2022). IEEE

    2022

  77. [77]

    In: 2022 ACM/IEEE 13th International Conference on Cyber-Physical Systems (ICCPS), pp

    Malawade, A.V., Mortlock, T., Al Faruque, M.A.: Hydrafusion: Context-aware selec- tive sensor fusion for robust and efficient autonomous vehicle perception. In: 2022 ACM/IEEE 13th International Conference on Cyber-Physical Systems (ICCPS), pp. 68–79 (2022). IEEE 42

    2022

  78. [78]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Chen, X., Zhang, T., Wang, Y., Wang, Y., Zhao, H.: Futr3d: A unified sensor fusion framework for 3d detection. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 172–181 (2023)

    2023

  79. [79]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Wu, P., Chen, S., Metaxas, D.N.: Motionnet: Joint perception and motion prediction for autonomous driving based on bird’s eye view maps. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 11385–11395 (2020)

    2020

  80. [80]

    arXiv preprint arXiv:2205.09743 (2022)

    Zhang, Y., Zhu, Z., Zheng, W., Huang, J., Huang, G., Zhou, J., Lu, J.: Beverse: Unified perception and prediction in birds-eye-view for vision-centric autonomous driving. arXiv preprint arXiv:2205.09743 (2022)

    Pith/arXiv arXiv 2022

Showing first 80 references.

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.