Defer to Plan: Adaptive Multi-Agent Fusion for End-to-End V2X Driving

Chao Sun; Nuoran Li; Tianze Wang; Yueran Zhao; Zhang Zhang

REVIEW 4 major objections 7 minor 24 references

Defer to Plan: Adaptive Multi-Agent Fusion for End-to-End V2X Driving

T0 review · 4 major / 7 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read Deferring multi-agent fusion to the planning stage lifts V2X driving score by 3.33% over the prior best.

desk verdict Planning-stage adaptive fusion is a real design idea, but the key ablation doesn't isolate it from decoder capacity, and the closed-loop numbers lack error bars. read the letter →

arxiv 2607.19774 v1 pith:E6ZPG7GD submitted 2026-07-22 cs.RO

Nuoran Li , Zhang Zhang , Yueran Zhao , Tianze Wang , Chao Sun This is my paper · ORCID

classification cs.RO

keywords V2Xdrivingcollaborativeend-to-endplanningautoregressivedecodermulti-agentfusionmixture-of-expertsclosed-loopevaluationtrajectoryprediction

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 argues that in vehicle-to-everything (V2X) autonomous driving, fusing features from other vehicles and roadside sensors should happen at the planning stage, not the perception stage. The proposed system uses an autoregressive decoder to learn scenario-dependent weights for combining ego and road tokens, so the relative importance of roadside information can shift from an occluded intersection to an open highway. On a closed-loop benchmark, the method achieves a driving score of 79.72 versus 77.15 for the prior state-of-the-art, with fewer infractions and lower trajectory error. The author's core case is that fusion timing, not just fusion architecture, is the key design choice for end-to-end collaborative driving.

What carries the argument

The central mechanism is the autoregressive decoder with a hybrid causal mask. It concatenates a target embedding, ego tokens, road tokens, and planning queries into one sequence; context tokens attend to each other bidirectionally, while each planning waypoint attends to the full context and only preceding waypoints. This lets the decoder dynamically re-weight ego versus roadside information during trajectory generation, replacing fixed perception-stage fusion. Supporting machinery includes a dual-stream tokenizer with channel-wise top-10% purification of roadside features and a mixture-of-experts layer with trajectory-guided routing to prevent expert collapse.

What would settle it

Re-run the exact closed-loop benchmark protocol for both the proposed system and CoDriving with at least ten random seeds and report mean plus standard deviation; if the driving-score gap falls below the run-to-run variance, the claimed improvement is not established. Alternatively, freeze the decoder's fusion weights to fixed equal weights during evaluation: if the driving score stays essentially the same, the adaptive weighting is not doing the work.

Watch

Extended reading notes

Core claim

The central claim is that multi-agent feature fusion should be deferred to the planning stage, where an autoregressive decoder learns adaptive, scenario-dependent fusion weights, rather than being performed at the perception stage with fixed weights. The system combines channel-wise purification of roadside features, mixture-of-experts tokenization, and an autoregressive decoder with a hybrid causal mask to jointly reason over context tokens and generate waypoints. On the V2Xverse closed-loop benchmark, this design raises the driving score from 77.15 to 79.72 and the infraction score from 0.82 to 0.88, while also degrading less than the baseline under pose noise and communication latency.

Load-bearing premise

The central claim rests on the closed-loop comparison against the CoDriving baseline being conducted under identical evaluation conditions; the paper reports no seeds, rollout counts, or confidence intervals, so if the baseline numbers came from a different protocol or run, the 3.33% driving-score gap may not be real.

Editorial extensions

If this is right

If the claim holds, perception-stage fusion methods in V2X driving are leaving measurable performance on the table, even when their perception metrics look strong.
The reported infraction-score gain (0.82 to 0.88) suggests that planning-stage adaptive fusion produces safer trajectories with fewer traffic violations.
Robustness results under pose noise and latency indicate that adaptive re-weighting can compensate for degraded or stale collaborative information, which matters for real-world deployment.
Ablation studies attribute the largest single improvement to the autoregressive decoder, implying that fusion timing itself is a primary source of gain, not just feature processing.
The measured planning latency (18.4 ms total, 10.2 ms for the decoder) suggests the approach is compatible with real-time control on a single GPU.

Reading between the lines

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

A testable extension would be to replace the autoregressive decoder with a simpler per-scenario weighting function (for example, a small MLP conditioned on scene features) to isolate whether the gain comes from adaptivity or from the decoder's representational capacity.
The adaptive attention weights over ego versus road tokens could serve as an interpretability signal, letting engineers see which agent's information drove a braking or turning decision.
The same 'defer fusion to the planning stage' principle may generalize to single-vehicle sensor fusion, where camera and LiDAR features are still often fused early; planning-stage adaptive weighting could be tested there.
The paper's assumption of a fixed communication topology leaves open a dynamic version where the planner selects which agents to attend to per scenario, which could cut bandwidth further.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Quick take: the planning-stage adaptive fusion idea is real and worth taking seriously; the paper's own ablation doesn't cleanly prove it, and the closed-loop numbers come without error bars. Still, this is one of the more sensible V2X-AD papers I've read recently.

What's actually new: deferring multi-agent fusion to a planning-stage autoregressive decoder that learns scenario-dependent attention weights, instead of fixed perception-stage fusion. That's a meaningful design change, and the hybrid causal mask (bidirectional context, causal planning queries) is a sensible way to parallelize generation. The dual-stream tokenization with channel-wise purification and MoE routing guided by trajectory-curvature K-means is also new to this setting, and the expert activation analysis suggests the routing actually specializes. The paper is well-written and the relation to CoDriving is clear.

Where it gets soft: the stress-test concern is legitimate. Table V's "w/o Autoregressive Decoder" removes the AR decoder entirely, which changes both the fusion mechanism and the capacity/architecture of the trajectory head. The improvement credited to 'adaptive fusion' could just be a bigger, better planner. The missing control is the same AR decoder fed with perception-stage fused features. Without it, the central claim about fusion timing isn't isolated. The fact that the w/o-AR variant (0.628 ADE) is worse than the CoDriving baseline (0.619) also suggests their fixed-fusion implementation is weak, which further muddies attribution.

On top of that, the reader's point stands: no seeds, no error bars, no rollout counts, and it's unclear if the CoDriving DS=77.15 was re-run or copied. A 3.33% closed-loop gap can easily be run-to-run noise in CARLA-based benchmarks. Hyperparameters like α, expert count, λ0 are hand-set with no sensitivity analysis, and ablation is only reported on open-loop ADE/FDE, not DS/IS.

None of this kills the paper. The core design idea is plausible and the open-loop gains are directionally consistent. But the evidence as presented supports 'the full system works better than CoDriving' more than it supports 'deferring fusion to planning is the reason.' The authors should add the decoder-capacity control, re-run or at least clearly state the provenance of all baseline numbers, and report variance.

Bottom line: worth a serious referee, but as is I'd want substantial revision. If you work on V2X or end-to-end driving, read it for the design idea; don't quote the DS numbers yet.

Referee Report

4 major / 7 minor

Summary. The paper proposes an end-to-end V2X collaborative driving system that shifts multi-agent feature fusion from the perception stage to the planning stage. The architecture has dual-stream feature processing (ego and roadside), channel-wise purification of roadside features, MoE-enhanced tokenization, and a LLaMA-style autoregressive decoder that learns scenario-dependent fusion weights while generating waypoints. On the V2Xverse benchmark, the authors report a closed-loop driving score of 79.72 versus 77.15 for CoDriving, lower ADE/FDE, and better infraction score, with robustness to pose noise and communication latency.

Significance. If the central claim holds, the 'defer to plan' design principle would be a useful contribution: adaptive, planning-aware fusion at the trajectory-generation stage instead of fixed perception-stage fusion. The paper's strengths are its end-to-end formulation, explicit robustness experiments under noise and latency, and open-loop gains over several baselines. The work is also not circular: the main comparisons are against an external method on an external benchmark. However, the key evidence is currently undermined by a confounded ablation and an under-specified closed-loop evaluation protocol, so the significance of the claimed 3.33% DS improvement is not yet established.

major comments (4)

[§4-D, Table V] The ablation 'w/o Autoregressive Decoder' changes both the fusion stage (planning-stage adaptive fusion vs. perception-stage fixed fusion) and the planning head (a multi-layer transformer decoder vs. a presumably simpler head). The 4.85% ADE gain credited to deferring fusion could instead be due to decoder capacity. The necessary control — the same autoregressive decoder fed with perception-stage fused features — is missing. Moreover, the 'w/o AR Decoder' variant (0.628 ADE) is worse than the CoDriving baseline (0.619 ADE), suggesting the fixed-fusion ablation implementation is not a strong baseline. Without this control, the paper's core design principle is not cleanly isolated.
[§4-A, §4-B, Table II] The closed-loop evaluation protocol is underspecified. No seeds, number of evaluation runs, variance, or confidence intervals are reported, and it is not stated whether the CoDriving baseline numbers (DS=77.15, RC=92.34, IS=0.82) were re-run under identical conditions or taken from [4]. The reported gap of 2.57 DS points (3.33%) could be within run-to-run variability of the simulator or due to protocol differences (e.g., route selection, infraction definitions). Please provide full protocol details and repeated-run statistics to establish the closed-loop improvement.
[§4-D, Table V] The ablation study reports only open-loop ADE/FDE, but the paper's headline result is the closed-loop driving score (DS/IS/RC). Without closed-loop ablations, there is no evidence that the proposed components (MoE tokenizers, channel-wise purification, autoregressive fusion) improve the closed-loop metrics that constitute the main claim. Please add closed-loop DS/IS/RC for each ablation configuration, or justify why open-loop errors are sufficient for the central claim.
[§3-C2, §3-E, Table V] Key hyperparameters are hand-set without sensitivity analysis: α=0.1 (channel retention), λ0=0.03 (router guidance weight), expert count=6, Top-3 routing, K=6 (maneuver clusters). Because the method's components are justified by these choices, the reported gains may hinge on fragile parameter settings. Please provide sensitivity sweeps or a principled selection procedure for α, λ0, expert count, and K.

minor comments (7)

[Abstract, §4-C] The abstract claims 'maintaining communication efficiency,' but no bandwidth or communication cost metric is reported; the latency table (Table I) reflects inference time, not message size. Please specify what communication efficiency is claimed and how it is measured.
[Eq. (1)] The problem formulation introduces P^{j→i} and a bandwidth constraint ||P^{j→i}|| ≤ B, but the paper never maps this notation to actual message sizes or evaluates the constraint. Please connect the formulation to the experiments or remove the unmeasured constraint.
[Table II] Several baselines have missing entries: TCP has no ADE/FDE, V2X-ViT has no closed-loop DS/IS/RC, Coopernaut has no closed-loop entries. For fair comparison, either fill these entries or state why they are unavailable.
[§3-D2] The decoder is called 'LLaMA-Style Parallel Autoregressive Decoder,' but LLaMA is not typically described as parallel; the hybrid causal mask enables parallel decoding in this paper. Please clarify the terminology to avoid confusion.
[§3-C3] The claim that Top-3 routing among 6 experts 'reduces inference cost while maintaining model capacity' depends on how the MoE layer is implemented (e.g., whether experts are sequentially activated or batched). Please specify the implementation to substantiate the efficiency statement.
[§3-C3, §4-E] The K-means maneuver clustering procedure is described only as 'clustering on trajectory curvature.' Please provide details: how is curvature computed, what trajectories are clustered, and how the K=6 clusters correspond to the MoE expert count.
[Overall] There are minor typographical issues, e.g., 'two-stage paradigm' (roman type), 'anend-to-end' in the introduction, and inconsistent spacing. Please proofread.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central claim is learned end-to-end and evaluated against an external baseline, not reduced to fitted inputs or self-citations.

full rationale

The paper's load-bearing claim — that deferring multi-agent fusion to the planning stage via an autoregressive decoder improves driving performance — does not reduce by construction to its inputs. The adaptive fusion weights are learned by L1 waypoint regression (Eq. 5, Eq. 6) and evaluated on an external simulator (V2Xverse, Ref. [4]); the decisive comparison (DS 79.72 vs. CoDriving 77.15) is against an external method by different authors, so no fitted parameter is renamed as a prediction. The self-citations [5]–[7], [12]–[16] support roadside perception modules and are not used to produce the main result. The MoE routing uses K-means curvature labels from training trajectories as auxiliary regularization (Eq. 7), but that is not circular: the reported ADE/DS metrics are not defined in terms of those labels. The Table V ablation removing the autoregressive decoder does conflate fusion timing with planning-head capacity, which is a genuine experimental confound and a reproducibility/correctness risk, but it is not a definitional or self-citational circularity. Likewise, the possibility that CoDriving baseline numbers were copied from [4] rather than re-run is a measurement/tooling concern, not a circular-derivation concern. No equation or fitted constant forces the headline result, so the derivation chain is self-contained with respect to circularity.

Assumptions & free parameters 5 free parameters · 5 assumptions · 1 invented entities

The system's empirical claims rest on standard ML machinery plus a set of hand-set design choices (α, expert count, K, annealing) and two unvalidated proxies (mean channel activation, curvature clustering). No new physical entities are introduced; the maneuver prototypes are the only new latent construct.

free parameters (5)

Roadside purification retention α (top-k) = 0.1
Chosen by hand in §III-C2; no sensitivity sweep, yet gates all roadside features and is credited with 2.34% ADE gain.
MoE shape (experts, top-k routing) = 6 experts, Top-3
Set in §III-C3/Fig.3; no ablation over expert count or routing ratio.
K-means maneuver clusters K = 6
Router supervision labels in §III-E; K matches expert count and is not validated against alternative clusterings.
Router guidance schedule (λ0, T_anneal) = λ0=0.03; T_anneal not reported
Eq. 7; annealing schedule is hand-picked and T_anneal is left unspecified.
Architecture sizes (d, k_e, k_r, T_f, #decoder layers) = Not reported
Required to reproduce the model; omitted from §III-C/D.

assumptions (5)

standard math Transformer attention, MoE routing, and L1 trajectory regression work as described by [18], [20], [23].
Technical machinery of the decoder, tokenizers, and loss.
domain assumption Closed-loop V2Xverse simulator faithfully ranks driving competence.
All central numbers come from this simulator (§IV-A); no real-world validation.
ad hoc to paper Roadside BEV features are 75%-90% redundant background.
Uncited assertion used to justify top-k purification (§I, §III-C2).
ad hoc to paper Mean spatial channel activation proxies planning relevance.
Top-k by mean activation (§III-C2) is not derived or swept.
ad hoc to paper K-means curvature clusters are meaningful maneuver labels for router supervision.
Used in §III-E to prevent expert collapse; cluster meaningfulness is not independently tested.

invented entities (1)

Maneuver prototypes from trajectory-curvature K-means
purpose: Soft labels that supervise the MoE router to avoid expert/router collapse
Constructed from the training set's own trajectories (§III-E); no evidence outside this pipeline that the clusters correspond to driver-relevant maneuvers.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Defer to Plan: Adaptive Multi-Agent Fusion for End-to-End V2X Driving." pith.science (2026). https://pith.science/paper/E6ZPG7GD

@misc{pith2026260719774,
  author       = {Pith},
  title        = {Pith review of: Defer to Plan: Adaptive Multi-Agent Fusion for End-to-End V2X Driving},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/E6ZPG7GD}},
  note         = {Machine review of arXiv:2607.19774}
}

read the original abstract

Vehicle-to-everything-aided autonomous driving (V2X-AD) significantly enhances driving performance through information sharing. However, existing collaborative perception methods only optimize module-level perception capabilities and fail to effectively serve the ultimate planning and control tasks. We propose an end-to-end collaborative driving system that directly optimizes planning task performance. The system employs MotionNetwork to fuse historical temporal information, utilizes attention mechanisms to efficiently compress spatial features into compact tokens, and adaptively fuses multi-agent features through an autoregressive decoder. Additionally, we introduce Mixture-of-Experts (MoE) architecture to enhance the model's representation capacity for heterogeneous features. Experiments demonstrate that our method achieves a driving score of 79.72, surpassing the state-of-the-art CoDriving baseline (77.15) by 3.33% in closed-loop evaluation while maintaining communication efficiency.

Figures

Figures reproduced from arXiv: 2607.19774 by the authors.

**Figure 1.** Perception-stage fusion vs Planning-stage fusion. Existing methods [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗

**Figure 3.** Detailed pipeline. Top: Ego stream processes BEV features through MotionNetwork and attention pooling. Bottom: Road stream applies channel-wise [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: Trajectory visualization in a yield-and-turn scenario. When a cyclist [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

**Figure 5.** Figure 5: MoE expert activation patterns across four scenarios. Different [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]

**Figure 6.** Figure 6: Channel-wise purification effect on roadside BEV features. The [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

24 extracted references · 6 linked inside Pith

[4]

Towards collaborative autonomous driving: Simulation platform and end-to-end system,

Genjia Liu, Yue Hu, Chenxin Xu, Weibo Mao, Junhao Ge, Zhengzhong Huang, Yifan Jia, Yiming Sun, Yuntao Zhao, Jie Duan, et al., “Towards collaborative autonomous driving: Simulation platform and end-to-end system,”IEEE Transactions on Pattern Analysis and Machine Intelli- gence, 2025

2025
[1]

V2vnet: Vehicle-to-vehicle com- munication for joint perception and prediction,

Tsun-Hsuan Wang, Sivabalan Manivasagam, Ming Liang, Bin Yang, Wenyuan Zeng, and Raquel Urtasun, “V2vnet: Vehicle-to-vehicle com- munication for joint perception and prediction,” inEuropean Conference on Computer Vision. Springer, 2020, pp. 605–621

2020
[2]

V2x-vit: Vehicle-to-everything cooperative perception with vision transformer,

Runsheng Xu, Hao Xiang, Zhengzhong Tu, Xin Xia, Ming-Hsuan Yang, and Jiaqi Ma, “V2x-vit: Vehicle-to-everything cooperative perception with vision transformer,” inEuropean Conference on Computer Vision. Springer, 2022, pp. 107–124

2022
[3]

V2xpnp: Vehicle-to-everything spatio-temporal fusion for multi- agent perception and prediction,

Zewei Zhou, Hao Xiang, Zhaoliang Zheng, Seth Z Zhao, Mingyue Lei, Yun Zhang, Tianhui Cai, Xinyi Liu, Johnson Liu, Maheswari Bajji, et al., “V2xpnp: Vehicle-to-everything spatio-temporal fusion for multi- agent perception and prediction,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2025, pp. 25399–25409

2025
[5]

Heatv2x: Scalable heterogeneous collaborative perception via efficient alignment and interaction,

Yueran Zhao, Zhang Zhang, Chao Sun, Tianze Wang, Chao Yue, and Nuoran Li, “Heatv2x: Scalable heterogeneous collaborative perception via efficient alignment and interaction,”arXiv preprint arXiv:2511.10211, 2025

arXiv 2025
[6]

Height3d: A roadside visual framework based on height prediction in real 3-d space,

Zhang Zhang, Chao Sun, Bo Wang, Bin Guo, Da Wen, Tianyi Zhu, and Qili Ning, “Height3d: A roadside visual framework based on height prediction in real 3-d space,”IEEE Transactions on Intelligent Transportation Systems, 2025

2025
[7]

Pillarid: Rethinking backbone network designs for pillar-based 3d object detection in infras- tructure point cloud,

Zhang Zhang, Chao Sun, Bo Wang, and Da Wen, “Pillarid: Rethinking backbone network designs for pillar-based 3d object detection in infras- tructure point cloud,”IEEE Transactions on Intelligent Transportation Systems, vol. 27, no. 1, pp. 232–240, 2025

2025
[8]

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,” inProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, 2023, pp. 17853– 17862

2023

Show all 24 references

[9]

Multi-modal fu- sion transformer for end-to-end autonomous driving,

Aditya Prakash, Kashyap Chitta, and Andreas Geiger, “Multi-modal fu- sion transformer for end-to-end autonomous driving,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 7077–7087

2021
[10]

Trajectory-guided control prediction for end-to-end au- tonomous driving: A simple yet strong baseline,

Penghao Wu, Xiaosong Jia, Li Chen, Junchi Yan, Hongyang Li, and Yu Qiao, “Trajectory-guided control prediction for end-to-end au- tonomous driving: A simple yet strong baseline,”Advances in Neural Information Processing Systems, vol. 35, pp. 6119–6132, 2022

2022
[11]

End-to-end autonomous driving through v2x cooperation,

Haibao Yu, Wenxian Yang, Jiaru Zhong, Zhenwei Yang, Siqi Fan, Ping Luo, and Zaiqing Nie, “End-to-end autonomous driving through v2x cooperation,” inProceedings of the AAAI Conference on Artificial Intelligence, 2025, vol. 39, pp. 9598–9606

2025
[12]

Heightformer: Learning height prediction in voxel features for roadside vision centric 3d object detection via transformer,

Zhang Zhang, Chao Sun, Chao Yue, Da Wen, Yujie Chen, Tianze Wang, and Jianghao Leng, “Heightformer: Learning height prediction in voxel features for roadside vision centric 3d object detection via transformer,” arXiv preprint arXiv:2503.10777, 2025

2025 arXiv
[13]

Pillarmamba: Learning local-global context for roadside point cloud via hybrid state space model,

Zhang Zhang, Chao Sun, Chao Yue, Da Wen, Tianze Wang, and Jianghao Leng, “Pillarmamba: Learning local-global context for roadside point cloud via hybrid state space model,”arXiv preprint arXiv:2505.05397, 2025

2025 arXiv
[14]

Roadformer: Local-global feature fusion for road surface classification in autonomous driving,

Tianze Wang, Zhang Zhang, Chao Yue, Yueran Zhao, and Chao Sun, “Roadformer: Local-global feature fusion for road surface classification in autonomous driving,”IEEE Access, vol. 13, pp. 201875–201888, 2025

2025
[15]

Roadmamba: A dual branch visual state space model for road surface classification,

Tianze Wang, Zhang Zhang, Chao Yue, Nuoran Li, and Chao Sun, “Roadmamba: A dual branch visual state space model for road surface classification,”arXiv preprint arXiv:2508.01210, 2025

2025 arXiv
[16]

Lanemapnet: Lane network recognization and hd map construction using curve region aware temporal bird’s-eye-view perception,

Tianyi Zhu, Jianghao Leng, Jiaru Zhong, Zhang Zhang, and Chao Sun, “Lanemapnet: Lane network recognization and hd map construction using curve region aware temporal bird’s-eye-view perception,” in2024 IEEE Intelligent Vehicles Symposium (IV). IEEE, 2024, pp. 2168–2175

2024
[17]

Cooper- naut: End-to-end driving with cooperative perception for networked vehicles,

Jiaxun Cui, Hang Qiu, Dian Chen, Peter Stone, and Yuke Zhu, “Cooper- naut: End-to-end driving with cooperative perception for networked vehicles,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 17252–17262

2022
[18]

Outrageously large neural networks: The sparsely-gated mixture-of-experts layer,

Noam Shazeer, Azalia Mirhoseini, Krzysztof Maziarz, Andy Davis, Quoc Le, Geoffrey Hinton, and Jeff Dean, “Outrageously large neural networks: The sparsely-gated mixture-of-experts layer,”arXiv preprint arXiv:1701.06538, 2017

2017 arXiv
[19]

Drivemoe: Mixture-of-experts for vision-language-action model in end-to-end autonomous driving,

Zhenjie Yang, Yilin Chai, Xiaosong Jia, Qifeng Li, Yuqian Shao, Xuekai Zhu, Haisheng Su, and Junchi Yan, “Drivemoe: Mixture-of-experts for vision-language-action model in end-to-end autonomous driving,”arXiv preprint arXiv:2505.16278, 2025

2025 arXiv
[20]

Llama: Open and efficient foundation language models,

Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie- Anne Lachaux, Timoth´ee Lacroix, Baptiste Rozi `ere, Naman Goyal, Eric Hambro, Faisal Azhar, et al., “Llama: Open and efficient foundation language models,”arXiv preprint arXiv:2302.13971, 2023

2023 arXiv
[21]

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,” inEuropean Conference on Computer Vision. Springer, 2020, pp. 194–210

2020
[22]

Bevformer: learning bird’s-eye-view representation from lidar-camera via spatiotemporal transformers,

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 Transactions on Pattern Analysis and Machine Intelligence, 2024

2024
[23]

Attention is all you need,

Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin, “Attention is all you need,”Advances in neural information processing systems, vol. 30, 2017

2017
[24]

Tumtraf v2x cooperative perception dataset,

Walter Zimmer, Gerhard Arya Wardana, Suren Sritharan, Xingcheng Zhou, Rui Song, and Alois C. Knoll, “Tumtraf v2x cooperative perception dataset,” 2024, CoopDet3D is proposed in this paper

2024

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.

[1] [4]

Towards collaborative autonomous driving: Simulation platform and end-to-end system,

Genjia Liu, Yue Hu, Chenxin Xu, Weibo Mao, Junhao Ge, Zhengzhong Huang, Yifan Jia, Yiming Sun, Yuntao Zhao, Jie Duan, et al., “Towards collaborative autonomous driving: Simulation platform and end-to-end system,”IEEE Transactions on Pattern Analysis and Machine Intelli- gence, 2025

2025

[2] [1]

V2vnet: Vehicle-to-vehicle com- munication for joint perception and prediction,

Tsun-Hsuan Wang, Sivabalan Manivasagam, Ming Liang, Bin Yang, Wenyuan Zeng, and Raquel Urtasun, “V2vnet: Vehicle-to-vehicle com- munication for joint perception and prediction,” inEuropean Conference on Computer Vision. Springer, 2020, pp. 605–621

2020

[3] [2]

V2x-vit: Vehicle-to-everything cooperative perception with vision transformer,

Runsheng Xu, Hao Xiang, Zhengzhong Tu, Xin Xia, Ming-Hsuan Yang, and Jiaqi Ma, “V2x-vit: Vehicle-to-everything cooperative perception with vision transformer,” inEuropean Conference on Computer Vision. Springer, 2022, pp. 107–124

2022

[4] [3]

V2xpnp: Vehicle-to-everything spatio-temporal fusion for multi- agent perception and prediction,

Zewei Zhou, Hao Xiang, Zhaoliang Zheng, Seth Z Zhao, Mingyue Lei, Yun Zhang, Tianhui Cai, Xinyi Liu, Johnson Liu, Maheswari Bajji, et al., “V2xpnp: Vehicle-to-everything spatio-temporal fusion for multi- agent perception and prediction,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2025, pp. 25399–25409

2025

[5] [5]

Heatv2x: Scalable heterogeneous collaborative perception via efficient alignment and interaction,

Yueran Zhao, Zhang Zhang, Chao Sun, Tianze Wang, Chao Yue, and Nuoran Li, “Heatv2x: Scalable heterogeneous collaborative perception via efficient alignment and interaction,”arXiv preprint arXiv:2511.10211, 2025

arXiv 2025

[6] [6]

Height3d: A roadside visual framework based on height prediction in real 3-d space,

Zhang Zhang, Chao Sun, Bo Wang, Bin Guo, Da Wen, Tianyi Zhu, and Qili Ning, “Height3d: A roadside visual framework based on height prediction in real 3-d space,”IEEE Transactions on Intelligent Transportation Systems, 2025

2025

[7] [7]

Pillarid: Rethinking backbone network designs for pillar-based 3d object detection in infras- tructure point cloud,

Zhang Zhang, Chao Sun, Bo Wang, and Da Wen, “Pillarid: Rethinking backbone network designs for pillar-based 3d object detection in infras- tructure point cloud,”IEEE Transactions on Intelligent Transportation Systems, vol. 27, no. 1, pp. 232–240, 2025

2025

[8] [8]

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,” inProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, 2023, pp. 17853– 17862

2023