Pith. sign in

REVIEW 3 major objections 5 minor 66 references

Radar4D-VLM: Proposal-Grounded Temporal 4D Radar Reasoning Across Frozen Language Models

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Radar-only tokens feed frozen LLMs; aligned text adds no gain

desk verdict A carefully controlled empirical audit of a radar-only token interface for frozen LLMs; the language-supervision null holds up on its own terms, but the lack of an untouched test split and undisclosed checkpoint-selection metric keep it conditional. read the letter →

arxiv 2608.04130 v1 pith:OV5M4YIU submitted 2026-08-04 cs.CV

Jiaju Han , Xuemeng Sun , Qike Zhang , Xiang Chen , Luwei Yang , Jiahuan Long , Yiwei Wei , Jiujiang Guo
show 1 more author
Chengyin Hu
This is my paper · ORCID
classification cs.CV
keywords 4Dradarradar-onlyperceptionvision-languagemodelfrozenlanguagebackbonetemporalpointcloudreasoningautonomousdrivingcontrolledattributionobjectproposalgrounding
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

Radar4D-VLM argues that ten consecutive 4D-radar sweeps, organized into 64 object tokens, four scene tokens, and one kinematic token, are enough for frozen large language models to reason about moving objects, collision risk, and velocity without camera or LiDAR input. The paper's central claim is that this 69-token interface is compatible with every frozen backbone it audited, and that the language model consumes real radar evidence and temporal order. At the same time, it claims that adding aligned language supervision does not improve the shared radar representation: matched aligned, permuted, and no-language training objectives produce statistically indistinguishable direct-head accuracy. If correct, the result separates two things usually conflated in radar–language systems: whether a frozen LLM can consume radar tokens (it can) and whether language supervision helps radar perception (in this setting, no stable help was observed). This matters for autonomous-driving perception because radar is weather-robust, and the finding suggests a cheaper, auditable path to radar-only scene and motion prediction while warning that fluent language outputs alone do not prove that language objectives are earning their cost.

What carries the argument

The load-bearing object is the 69×256 radar state $H_t = [o_{t,1:64}; k_t; s_{t,1:4}]$, built from ten sweeps: 64 proposal-grounded object tokens, one kinematic token, and four scene tokens. A frozen RTNH-compatible sparse-convolutional encoder supplies Top-64 proposal centers and features; trainable tokenizers add temporal Doppler descriptors and scene context; a learned state query produces a representation $z_t$ that drives six direct prediction heads and also feeds a budgeted low-rank projector (under 1.2 million trainable parameters) into frozen language backbones. The controlled-attribution design is equally central: aligned, fixed-permutation, and no-language objectives share the same radar state, direct heads, windows, and optimization budget, so any difference in direct-head accuracy is attributable to the language objective rather than to interface compatibility.

What would settle it

Evaluate the same matched aligned/permuted/no-language audit on a locked holdout split of K-Radar (or another 4D-radar dataset) with more than three seeds and randomly drawn verbalizer permutations; if aligned supervision shows a stable positive direct-head effect with intervals excluding zero, the paper's bounded null fails. Separately, re-run the Top-64 proposal recall on the holdout; a recall drop toward the lattice control would show that the proposal-geometry claim is split-specific.

Watch

Extended reading notes

Core claim

The paper's central discovery is a bounded non-observation paired with a positive interface result. On K-Radar development validation, the learned Top-64 proposal centers reach 98.13% recall at 4 m, beating fixed-lattice (91.73%) and uniform-random (75.30%) controls, so proposal geometry carries real target information. The 69-token radar state, compressed through a low-rank projector, yields language-path core balanced accuracy in a narrow 0.4860–0.4995 range across eight frozen Qwen, Phi, Mistral, Llama, and Gemma models, establishing cross-family compatibility. The aligned language objective, however, shows no stable direct-head gain over a fixed verbalizer permutation (−0.0052) or no language at all (−0.0024), with crossed seed–sequence 95% intervals spanning zero; the paper calls this a bounded non-observation rather than evidence of equivalence. Sensor interventions—zeroing radar, shuffling windows, and reversing history—consistently reduce accuracy on both language and direct paths, so the aligned checkpoints are genuinely using radar content and temporal order. The paper's conclusion is that interface compatibility and sensor dependence are established, while benefit from aligned language supervision is not.

Load-bearing premise

The results all come from one development-validation split (K-Radar sequences 41–48, including the fixed 1,024-window subset used for checkpoint selection); because no untouched test split is evaluated, the reported effect sizes and the language-supervision null could be distorted by selection.

Editorial extensions

If this is right

  • Frozen language models can consume temporal 4D-radar evidence as a compact token sequence, so radar-only reasoning is feasible without camera or LiDAR inputs.
  • Aligned language supervision does not, in this setting, improve the shared radar representation; researchers should not assume caption-style objectives help perception heads.
  • Temporal history (ten sweeps versus one) provides the strongest consistent signal, larger than the Doppler channel, so multi-sweep accumulation is the priority for radar token design.
  • The pruned no-language exports retain direct-head accuracy (0.5014 mean balanced accuracy) without any language model, so radar-only predictors can be deployed without the language stack.
  • Reporting interface compatibility, sensor dependence, and language-supervision benefit as separate claims prevents fluent answers from being mistaken for grounded perception.

Reading between the lines

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

  • If the null holds on untouched test data, radar-perception research can invest in temporal tokenization and direct heads rather than in language-model alignment, which is cheaper and easier to audit.
  • The descriptive verbalizer sensitivity (mean range 0.0196 balanced-accuracy units across fixed permutations, with one pairwise interval excluding zero) suggests that answer-string mappings can shift results enough to demand random-permutation inference before concluding any supervision effect.
  • A direct extension would test the same matched audit on an unseen dataset or on K-Radar sequences 49–58 with a frozen protocol, checking whether the bounded null and the 98% proposal recall replicate outside the development split.
  • The compatibility result across eight backbones suggests that token structure, not backbone capacity, is the bottleneck for radar-language reasoning; comparing different token hierarchies under the same frozen backbone would isolate which structure matters.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. Radar4D-VLM proposes a radar-only temporal vision-language model that converts ten consecutive 4D-radar sweeps into a compact hierarchy of 64 object, four scene, and one kinematic token. A frozen RTNH-based encoder supplies Top-64 proposal centers; the 69 tokens feed both auditable direct prediction heads and a low-rank projector into frozen Qwen, Phi, Mistral, Llama, and Gemma backbones. On K-Radar sequences 41–48 (development validation), the paper reports 98.13% Top-64 proposal recall at 4 m, language-path core balanced accuracy between 0.4860 and 0.4995 across eight backbones, and controlled input interventions showing sensitivity to real radar and temporal order. The central attribution result is that matched aligned, fixed-permutation, and no-language objectives produce no stable gain in direct-head core balanced accuracy from aligned language supervision, with equal-sequence paired effects of −0.0052 and −0.0024 and crossed 95% intervals spanning zero. The paper explicitly frames all results as development-validation evidence, not held-out test performance.

Significance. The paper's main strength is its controlled-attribution design: it separates interface compatibility (frozen backbones consume the tokens), sensor dependence (interventions degrade outputs), and supervision benefit (direct-head comparisons across matched objectives). The equal-sequence paired bootstrap, sequence-weighted analysis, explicit disclosure of development-validation status, and candid limitation statements are exemplary. The negative language-supervision result is a falsifiable and non-obvious finding that challenges the common assumption that aligned language objectives improve shared radar representations. The proposal-versus-grid equal-count control and the objective-robust sensor dependence checks provide useful evidence. If the checkpoint-selection confound is resolved, the paper would be a solid empirical contribution to radar-language research.

major comments (3)
  1. [Data Isolation and Evaluation Manifest; Table 4] The checkpoint-selection rule for the trained direct-head/projector models is not reported. The text states that checkpoints are selected on a fixed 1,024-window subset of the 4,208 validation windows, and the proposal encoder is selected by Top-64 recall at 4 m, but no selection metric is given for the adapter and direct heads. If selection minimizes the total loss in Eq. (5), then aligned runs are selected jointly for language fluency, while permuted and no-language runs are selected only for direct and proposal losses; this would bias the aligned direct-head endpoint downward and make the Table 2 null at least partly an artifact of the selection rule. Because the language-value null is the central claim, the authors must disclose the selection metric for each condition and demonstrate that it is condition-independent (e.g., a fixed epoch or a direct-head-only criterion), or repeat the audit under a common selection rule.
  2. [Proposal geometry and representation structure; Table 6] The headline 98.13% Top-64 recall at 4 m is the value of the checkpoint selected by that same metric on the same development-validation sequences. This number is therefore an in-sample selection maximum, not an unbiased estimate; the abstract and the proposal-geometry section should explicitly state this, and a cross-validated or untouched-split estimate should be provided if an absolute claim is intended. The relative comparison against fixed-lattice and random proposals remains informative, but the absolute magnitude should not be reported without this caveat.
  3. [Matched Language-Supervision Audit; supplement 'Robustness to Additional Fixed Verbalizers'] The primary aligned-versus-permuted contrast uses a single fixed within-task derangement. The supplementary analysis of three additional derangements shows a mean direct-head BA range of 0.0196, with the P2−P3 pairwise interval excluding zero, indicating that the specific permutation can affect direct-head scores by an amount comparable to the paired effect in Table 2. The aligned-versus-no-language contrast is not subject to this issue, but the paper should either present that contrast as the primary test or average over a set of permutations and account for the mapping-induced variance in the intervals. As it stands, the aligned-versus-permuted null in Table 2 is not robust to the choice of derangement.
minor comments (5)
  1. [Abstract] The abstract's 'proposal recall reaches 98.13%' should be qualified as a development-validation selection result; consider adding a caveat or moving the number to the body.
  2. [Data Isolation and Evaluation Manifest] Clarify whether the 1,024-window checkpoint-selection subset is fully contained in the 4,208-window denominator, and whether the reported results are on the full 4,208 windows or on a subset.
  3. [Table 2] Specify the number of bootstrap samples and the exact resampling procedure for the crossed seed–sequence intervals, as is done for the supplement's sequence-cluster intervals.
  4. [Figure 4(a)] The Holm-adjusted p=1.000 is reported, but since an exact two-sided seed sign-flip test cannot attain conventional significance with three seeds, reporting p-values may be misleading even with the disclaimer; consider omitting them or presenting only the intervals.
  5. [Abstract and Conclusion] The phrase 'no stable direct-head gain' should be harmonized with the paper's own description as a bounded non-observation, for example by writing 'no stable direct-head gain was observed in this development-validation setting.'

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: headline results are empirical measurements with explicit controls and disclosed claim boundaries.

full rationale

The paper's central claims are direct empirical measurements rather than derivations from their own inputs. The 69-token interface compatibility result is measured across eight frozen backbones from five LLM families; the language-value null is measured via paired aligned/permuted/no-language objectives with direct-head endpoints evaluated independently of answer-token agreement; radar dependence is measured via input interventions; and proposal recall is measured against fixed-lattice and uniform-random controls. None of these numbers is defined in terms of the quantity it is used to support. The closest design concern is that all checkpoints are selected on the 1,024-window development-validation subset, and all reported results are development-validation results rather than untouched-test results. This is a data-selection and generalization concern, and the paper itself discloses it explicitly ('All reported results are explicitly development-validation results. No official nuScenes... or locked-holdout evaluation is claimed.'), but it is not circularity: the reported effects are still computed from model outputs on fixed windows, not forced by the training objective or by a self-citation chain. There is no equation in which a predicted quantity is defined via the target quantity, no fitted parameter is renamed as a prediction, and no load-bearing claim rests on a self-citation. The bounded non-observation of language-supervision benefit is an empirical null result with explicitly stated intervals and limitations. Therefore no significant circularity is present.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

No physical entities are postulated. The kinematic token is a model component whose purpose is to summarize weighted motion statistics, and its effect is tested by ablation (removal yields 0.4829 vs 0.4889 language-path BA). For an empirical ML paper, the ledger mainly records design choices and dataset assumptions rather than derived constants.

free parameters (5)
  • Top-64 proposal count
    Chosen cap of 64 proposal centers per frame; proposal recall (98.13% at 4 m) and the 69-token state depend on this hand-set number.
  • 10-sweep temporal window
    The temporal history length is fixed to ten sweeps; ablations show it provides the strongest signal, but the value is a design choice.
  • Token counts (4 scene, 1 kinematic)
    Number of scene queries and kinematic tokens is chosen by the authors; the 69-token interface is defined by these counts.
  • 4 m proposal recall radius = 4 m
    Evaluation threshold for Top-K recall; headline 98.13% recall is measured at this radius, and the grid/random comparisons use the same threshold.
  • Projector trainable budget = <=1.2M
    The matched budget ceiling across backbones is a hand-set constraint; it ensures fair comparison but is not derived.
assumptions (4)
  • domain assumption K-Radar track-derived labels correctly define the six task ontologies (moving count, sector, motion, collision risk, category, speed).
    Used in Section 'Task and Evaluation Endpoints' as supervision and evaluation; if labels are noisy, all scores shift.
  • domain assumption Top-64 proposal centers, trained on K-Radar sequences 1-40 and selected on 41-48, provide sufficient object-level geometry for downstream tokenization.
    The whole token hierarchy is proposal-grounded (Eq. 2); the paper measures recall but not precision, and assumes center proximity to targets is enough.
  • domain assumption Scene-local shuffle maps each window to another window from the same sequence without introducing confounds beyond scene identity.
    Used in Supplement 'Radar Information Controls Across Objectives'; if windows in a sequence are not exchangeable, the control is imperfect.
  • domain assumption Crossed seed-sequence bootstrap intervals with three seeds provide valid descriptive uncertainty for paired effects.
    The paper explicitly labels the audit as descriptive and notes an exact sign-flip test cannot reach significance; the bootstrap is used as a descriptive tool.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Radar4D-VLM: Proposal-Grounded Temporal 4D Radar Reasoning Across Frozen Language Models." pith.science (2026). https://pith.science/paper/OV5M4YIU

@misc{pith2026260804130,
  author       = {Pith},
  title        = {Pith review of: Radar4D-VLM: Proposal-Grounded Temporal 4D Radar Reasoning Across Frozen Language Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OV5M4YIU}},
  note         = {Machine review of arXiv:2608.04130}
}
read the original abstract

Vision-language models for autonomous driving primarily rely on cameras and LiDAR, leaving 4D radar largely unexplored as a standalone perceptual modality despite its robustness to adverse visibility and direct measurement of radial velocity. We introduce Radar4D-VLM, a radar-only temporal vision-language model that reasons from ten consecutive 4D-radar point-cloud sweeps without camera or LiDAR input. Radar4D-VLM extracts geometrically grounded object proposals and organizes radar evidence into a compact hierarchy of object, scene, and kinematic tokens. A parameter-efficient projector maps these tokens into frozen language backbones, while auditable prediction heads jointly model object count, spatial distribution, motion state, collision risk, semantic category, and radial velocity. Radar4D-VLM combines proposal-grounded temporal object tokenization, global scene context, and explicit kinematic tokens within a unified frozen-backbone interface. On sequence-isolated K-Radar development validation, its Top-64 proposal recall reaches 98.13% at 4 m, exceeding fixed-lattice and uniform-random controls by 6.40 and 22.83 percentage points, respectively. We further evaluate 24 matched runs spanning eight frozen Qwen, Phi, Mistral, Llama, and Gemma backbones under an identical adaptation budget. The radar-token interface remains compatible across all five language-model families, while matched aligned, permuted, and no-language controls show sensor dependence but no stable direct-head gain from aligned language supervision. These results establish a reproducible foundation for radar-only multimodal scene and motion reasoning while separating interface compatibility from the benefit of language supervision.

Figures

Figures reproduced from arXiv: 2608.04130 by the authors.

Figure 1
Figure 1. Radar4D-VLM architecture and controlled training conditions. The current sweep is processed by a frozen, train-split [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Proposal-to-return association in fog and snow. RGB is synchronized context only and is not a model input. [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Grounded Q&A cases spanning correct, partial, [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Controlled attribution on K-Radar development [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

66 extracted references · 60 canonical work pages

  1. [1]

    H.; Vora, S.; Liong, V

    Caesar, H.; Bankiti, V.; Lang, A. H.; Vora, S.; Liong, V. E.; Xu, Q.; Krishnan, A.; Pan, Y.; Baldan, G.; and Beijbom, O. 2020. nuScenes : A Multimodal Dataset for Autonomous Driving. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 11621--11631

    2020

  2. [2]

    Chae, Y.; Park, H.; Kim, H.; and Yoon, K.-J. 2025. Doppler-Aware LiDAR-RADAR Fusion for Weather-Robust 3D Detection. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 27197--27208

    work page 2025

  3. [3]

    Chen, S.; Chen, X.; Zhang, C.; Li, M.; Yu, G.; Fei, H.; Zhu, H.; Fan, J.; and Chen, T. 2024. LL3DA : Visual Interactive Instruction Tuning for Omni-3D Understanding, Reasoning, and Planning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 26428--26438

    work page 2024

  4. [4]

    M.; and Lu, C

    Ding, F.; Palffy, A.; Gavrila, D. M.; and Lu, C. X. 2023. Hidden Gems: 4D Radar Scene Flow Learning Using Cross-Modal Supervision. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 9340--9349

    work page 2023

  5. [5]

    Geirhos, R.; Jacobsen, J.-H.; Michaelis, C.; Zemel, R.; Brendel, W.; Bethge, M.; and Wichmann, F. A. 2020. Shortcut Learning in Deep Neural Networks. Nature Machine Intelligence, 2(11): 665--673

    2020

  6. [6]

    L.; Cai, X.; Xu, M.; Smith, J

    Guan, R.; Zhang, R.; Ouyang, N.; Liu, J.; Man, K. L.; Cai, X.; Xu, M.; Smith, J. S.; Lim, E. G.; Yue, Y.; and Xiong, H. 2025. Talk2Radar : Bridging Natural Language with 4D mmWave Radar for 3D Referring Expression Comprehension. In IEEE International Conference on Robotics and Automation, 10884--10891

    work page 2025

  7. [7]

    Haitman, Y.; and Bialer, O. 2025. DoppDrive : Doppler-Driven Temporal Aggregation for Improved Radar Object Detection. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 26085--26094

    work page 2025

  8. [8]

    Hamilton, K.; and Heckman, C. 2026. Weather-Robust Scene Semantics with Vision-Aligned 4D Radar. arXiv:2605.07367

    work page Pith review arXiv 2026

Show all 66 references
  1. [9]

    M.; and Porikli, F

    Hegde, D.; Yasarla, R.; Cai, H.; Han, S.; Bhattacharyya, A.; Mahajan, S.; Liu, L.; Garrepalli, R.; Patel, V. M.; and Porikli, F. 2025. Distilling Multi-Modal Large Language Models for Autonomous Driving. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern ...

    2025

  2. [10]

    Hong, Y.; Zhen, H.; Chen, P.; Zheng, S.; Du, Y.; Chen, Z.; and Gan, C. 2023. 3D-LLM : Injecting the 3D World into Large Language Models. In Advances in Neural Information Processing Systems, volume 36, 20482--20494

    2023

  3. [11]

    Huang, J.; Jia, B.; Wang, Y.; Zhu, Z.; Linghu, X.; Li, Q.; Zhu, S.-C.; and Huang, S. 2025 a . Unveiling the Mist over 3D Vision-Language Understanding: Object-Centric Evaluation with Chain-of-Analysis. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Re...

    2025

  4. [12]

    Huang, T.; Prabhakara, A.; Chen, C.; Karhade, J.; Ramanan, D.; O'Toole, M.; and Rowe, A. 2025 b . Towards Foundational Models for Single-Chip Radar. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 24655--24665

    2025

  5. [14]

    Kung, P.-C.; Harisha, S.; Vasudevan, R.; Eid, A.; and Skinner, K. A. 2025. RadarSplat : Radar Gaussian Splatting for High-Fidelity Data Synthesis and 3D Reconstruction of Autonomous Driving Scenes. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 275...

    2025

  6. [15]

    Lai, Z.; Yang, J.; Xia, S.; Lin, L.; Sun, L.; Wang, R.; Liu, J.; Wu, Q.; and Pei, L. 2026. RadarLLM : Empowering Large Language Models to Understand Human Motion from Millimeter-wave Point Cloud Sequence. Proceedings of the AAAI Conference on Artificial Intelligence, 40(7): 5791--5799

    2026

  7. [16]

    Li, J.; Li, D.; Savarese, S.; and Hoi, S. 2023. BLIP -2: Bootstrapping Language-Image Pre-Training with Frozen Image Encoders and Large Language Models. In Proceedings of the 40th International Conference on Machine Learning, volume 202 of Proceedings of Machine Learning Resea...

    2023

  8. [17]

    Li, Y.; Tian, M.; Lin, Z.; Zhu, J.; Zhu, D.; Liu, H.; Zhang, Y.; Xiong, Z.; and Zhao, X. 2025. Fine-Grained Evaluation of Large Vision-Language Models in Autonomous Driving. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 9431--9442

    2025

  9. [18]

    Mei, G.; Lin, W.; Riz, L.; Wu, Y.; Poiesi, F.; and Wang, Y. 2025. PerLA : Perceptive 3D Language Assistant. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 14369--14379

    2025

  10. [20]

    Paek, D.-H.; Kong, S.-H.; and Wijaya, K. T. 2022. K-Radar : 4D Radar Object Detection for Autonomous Driving in Various Weather Conditions. In Advances in Neural Information Processing Systems, volume 35, 3819--3829. Curran Associates, Inc. Datasets and Benchmarks Track

    2022

  11. [21]

    Paul, S.; Patterson, Z.; and Bouguila, N. 2026. Point Cloud as a Foreign Language for Multi-modal Large Language Model. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 16676--16687

    2026

  12. [22]

    Pushkareva, M.; Feldman, Y.; Domokos, C.; Rambach, K.; and Di Castro, D. 2024. Radar Spectra-Language Model for Automotive Scene Parsing. arXiv:2406.02158

    2024 arXiv

  13. [23]

    Qian, T.; Chen, J.; Zhuo, L.; Jiao, Y.; and Jiang, Y.-G. 2024. NuScenes-QA : A Multi-Modal Visual Question Answering Benchmark for Autonomous Driving Scenario. Proceedings of the AAAI Conference on Artificial Intelligence, 38(5): 4542--4550

    2024

  14. [24]

    Sima, C.; Renz, K.; Chitta, K.; Chen, L.; Zhang, H.; Xie, C.; Bei wenger, J.; Luo, P.; Geiger, A.; and Li, H. 2024. DriveLM : Driving with Graph Visual Question Answering. In Computer Vision -- ECCV 2024, 256--274

    2024

  15. [25]

    Standley, T.; Zamir, A.; Chen, D.; Guibas, L.; Malik, J.; and Savarese, S. 2020. Which Tasks Should Be Learned Together in Multi-Task Learning? In Proceedings of the 37th International Conference on Machine Learning, volume 119 of Proceedings of Machine Learning Research, 9120--9132

    2020

  16. [26]

    Tong, T.; Liu, X.; Liu, H.; and Fan, B. 2026. Towards Accurate 3D Object Detection in Adverse Weather by Leveraging 4D Radar for LiDAR Geometry Enhancement. Proceedings of the AAAI Conference on Artificial Intelligence, 40(12): 9548--9556

    2026

  17. [27]

    Wang, J.; Yu, J.; Lin, H.; Liu, H.; Zhu, W.; and Xiong, S. 2026. Driving with Advice: Large Model as Motion Advisor for Joint Planning. Proceedings of the AAAI Conference on Artificial Intelligence, 40(2): 1168--1176

    2026

  18. [28]

    A.; Liu, Z.; and Pan, L

    Xie, S.; Kong, L.; Dong, Y.; Sima, C.; Zhang, W.; Chen, Q. A.; Liu, Z.; and Pan, L. 2025. Are VLMs Ready for Autonomous Driving? An Empirical Study from the Reliability, Data, and Metric Perspectives. In Proceedings of the IEEE/CVF International Conference on Computer Vision, ...

    2025

  19. [29]

    Xu, R.; Xiang, Z.; Zhang, C.; Zhong, H.; Zhao, X.; Dang, R.; Xu, P.; Pu, T.; and Liu, E. 2025. SCKD : Semi-Supervised Cross-Modality Knowledge Distillation for 4D Radar Object Detection. Proceedings of the AAAI Conference on Artificial Intelligence, 39(9): 8933--8941

    2025

  20. [30]

    F.; and Chai, J

    Yang, J.; Chen, X.; Madaan, N.; Iyengar, M.; Qian, S.; Fouhey, D. F.; and Chai, J. 2025. 3D-GRAND : A Million-Scale Dataset for 3D-LLMs with Better Grounding and Less Hallucination. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 29501--29512

    2025

  21. [31]

    Yuan, Z.; Peng, Y.; Ren, J.; Liao, Y.; Han, Y.; Feng, C.-M.; Zhao, H.; Li, G.; Cui, S.; and Li, Z. 2025. Empowering Large Language Models with 3D Situation Awareness. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 19435--19445

    2025

  22. [32]

    Zheng, L.; Ma, Z.; Zhu, X.; Tan, B.; Li, S.; Long, K.; Sun, W.; Chen, S.; Zhang, L.; Wan, M.; Huang, L.; and Bai, J. 2022. TJ4DRadSet : A 4D Radar Dataset for Autonomous Driving. In 2022 IEEE 25th International Conference on Intelligent Transportation Systems, 493--498

    2022

  23. [33]

    Zhong, H.; Xiang, Z.; Xu, R.; Fu, J.; Xu, P.; Wang, S.; Yang, Z.; Pu, T.; and Liu, E. 2025. CVFusion : Cross-View Fusion of 4D Radar and Camera for 3D Object Detection. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 28188--28197

    2025

  24. [34]

    Zhou, X.; Han, X.; Yang, F.; Ma, Y.; Tresp, V.; and Knoll, A. 2026. OpenDriveVLA : Towards End-to-End Autonomous Driving with Large Vision Language Action Model. Proceedings of the AAAI Conference on Artificial Intelligence, 40(16): 13782--13790

    2026

  25. [35]

    2022 , doi =

    Paek, Dong-Hee and Kong, Seung-Hyun and Wijaya, Kevin Tirta , booktitle =. 2022 , doi =

    2022

  26. [36]

    and Vora, Sourabh and Liong, Venice Erin and Xu, Qiang and Krishnan, Anush and Pan, Yu and Baldan, Giancarlo and Beijbom, Oscar , booktitle =

    Caesar, Holger and Bankiti, Varun and Lang, Alex H. and Vora, Sourabh and Liong, Venice Erin and Xu, Qiang and Krishnan, Anush and Pan, Yu and Baldan, Giancarlo and Beijbom, Oscar , booktitle =. 2020 , doi =

    2020

  27. [37]

    2022 , doi =

    Zheng, Lianqing and Ma, Zhixiong and Zhu, Xichan and Tan, Bin and Li, Sen and Long, Kai and Sun, Weiqi and Chen, Sihan and Zhang, Lu and Wan, Mengyue and Huang, Libo and Bai, Jie , booktitle =. 2022 , doi =

    2022

  28. [38]

    2310.17659 , archiveprefix =

    Kong, Seung-Hyun and Paek, Dong-Hee and Cho, Sangjae , year =. 2310.17659 , archiveprefix =

    arXiv

  29. [39]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages =

    Hidden Gems: 4D Radar Scene Flow Learning Using Cross-Modal Supervision , author =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages =. 2023 , doi =

    2023

  30. [40]

    and Lim, Eng Gee and Yue, Yutao and Xiong, Hui , booktitle =

    Guan, Runwei and Zhang, Ruixiao and Ouyang, Ningwei and Liu, Jianan and Man, Ka Lok and Cai, Xiaohao and Xu, Ming and Smith, Jeremy S. and Lim, Eng Gee and Yue, Yutao and Xiong, Hui , booktitle =. 2025 , doi =

    2025

  31. [41]

    2024 , eprint =

    Radar Spectra-Language Model for Automotive Scene Parsing , author =. 2024 , eprint =

    2024

  32. [42]

    2511.21105 , archiveprefix =

    Mishra, Pushkal and Bansal, Kshitiz and Bharadia, Dinesh , year =. 2511.21105 , archiveprefix =

  33. [43]

    2026 , eprint =

    Weather-Robust Scene Semantics with Vision-Aligned 4D Radar , author =. 2026 , eprint =

    2026

  34. [44]

    2026 , doi =

    Lai, Zengyuan and Yang, Jiarui and Xia, Songpengcheng and Lin, Lizhou and Sun, Lan and Wang, Renwen and Liu, Jianran and Wu, Qi and Pei, Ling , journal =. 2026 , doi =

    2026

  35. [45]

    2024 , doi =

    Qian, Tianwen and Chen, Jingjing and Zhuo, Linhai and Jiao, Yang and Jiang, Yu-Gang , journal =. 2024 , doi =

    2024

  36. [46]

    Computer Vision -- ECCV 2024 , pages =

    Sima, Chonghao and Renz, Katrin and Chitta, Kashyap and Chen, Li and Zhang, Hanxue and Xie, Chengen and Bei. Computer Vision -- ECCV 2024 , pages =. 2024 , doi =

    2024

  37. [47]

    2023 , url =

    Li, Junnan and Li, Dongxu and Savarese, Silvio and Hoi, Steven , booktitle =. 2023 , url =

    2023

  38. [48]

    2023 , doi =

    Hong, Yining and Zhen, Haoyu and Chen, Peihao and Zheng, Shuhong and Du, Yilun and Chen, Zhenfang and Gan, Chuang , booktitle =. 2023 , doi =

    2023

  39. [49]

    2024 , doi =

    Chen, Sijin and Chen, Xin and Zhang, Chi and Li, Mingsheng and Yu, Gang and Fei, Hao and Zhu, Hongyuan and Fan, Jiayuan and Chen, Tao , booktitle =. 2024 , doi =

    2024

  40. [50]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages =

    Point Cloud as a Foreign Language for Multi-modal Large Language Model , author =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages =. 2026 , url =

    2026

  41. [51]

    Proceedings of the 37th International Conference on Machine Learning , series =

    Which Tasks Should Be Learned Together in Multi-Task Learning? , author =. Proceedings of the 37th International Conference on Machine Learning , series =. 2020 , url =

    2020

  42. [52]

    Nature Machine Intelligence , volume =

    Shortcut Learning in Deep Neural Networks , author =. Nature Machine Intelligence , volume =. 2020 , doi =

    2020

  43. [53]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages =

    Towards Foundational Models for Single-Chip Radar , author =. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages =. 2025 , doi =

    2025

  44. [54]

    2025 , doi =

    Haitman, Yuval and Bialer, Oded , booktitle =. 2025 , doi =

    2025

  45. [55]

    2025 , doi =

    Zhong, Hanzhi and Xiang, Zhiyu and Xu, Ruoyu and Fu, Jingyun and Xu, Peng and Wang, Shaohong and Yang, Zhihao and Pu, Tianyu and Liu, Eryun , booktitle =. 2025 , doi =

    2025

  46. [56]

    , booktitle =

    Kung, Pou-Chun and Harisha, Skanda and Vasudevan, Ram and Eid, Aline and Skinner, Katherine A. , booktitle =. 2025 , doi =

    2025

  47. [57]

    Doppler-Aware

    Chae, Yujeong and Park, Heejun and Kim, Hyeonseong and Yoon, Kuk-Jin , booktitle =. Doppler-Aware. 2025 , doi =

    2025

  48. [58]

    2025 , doi =

    Xu, Ruoyu and Xiang, Zhiyu and Zhang, Chenwei and Zhong, Hanzhi and Zhao, Xijun and Dang, Ruina and Xu, Peng and Pu, Tianyu and Liu, Eryun , journal =. 2025 , doi =

    2025

  49. [59]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages =

    Distilling Multi-Modal Large Language Models for Autonomous Driving , author =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages =. 2025 , doi =

    2025

  50. [60]

    Xie, Shaoyuan and Kong, Lingdong and Dong, Yuhao and Sima, Chonghao and Zhang, Wenwei and Chen, Qi Alfred and Liu, Ziwei and Pan, Liang , booktitle =. Are. 2025 , doi =

    2025

  51. [61]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages =

    Fine-Grained Evaluation of Large Vision-Language Models in Autonomous Driving , author =. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages =. 2025 , doi =

    2025

  52. [62]

    2025 , doi =

    Mei, Guofeng and Lin, Wei and Riz, Luigi and Wu, Yujiao and Poiesi, Fabio and Wang, Yiming , booktitle =. 2025 , doi =

    2025

  53. [63]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages =

    Unveiling the Mist over 3D Vision-Language Understanding: Object-Centric Evaluation with Chain-of-Analysis , author =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages =. 2025 , doi =

    2025

  54. [64]

    and Chai, Joyce , booktitle =

    Yang, Jianing and Chen, Xuweiyi and Madaan, Nikhil and Iyengar, Madhavan and Qian, Shengyi and Fouhey, David F. and Chai, Joyce , booktitle =. 2025 , doi =

    2025

  55. [65]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages =

    Empowering Large Language Models with 3D Situation Awareness , author =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages =. 2025 , doi =

    2025

  56. [66]

    2026 , doi =

    Zhou, Xingcheng and Han, Xuyuan and Yang, Feng and Ma, Yunpu and Tresp, Volker and Knoll, Alois , journal =. 2026 , doi =

    2026

  57. [67]

    Towards Accurate 3D Object Detection in Adverse Weather by Leveraging 4D Radar for

    Tong, Tianxu and Liu, Xinrun and Liu, Hongmin and Fan, Bin , journal =. Towards Accurate 3D Object Detection in Adverse Weather by Leveraging 4D Radar for. 2026 , doi =

    2026

  58. [68]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume =

    Driving with Advice: Large Model as Motion Advisor for Joint Planning , author =. Proceedings of the AAAI Conference on Artificial Intelligence , volume =. 2026 , doi =

    2026

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.