Pith. sign in

REVIEW 2 major objections 6 minor 68 references

SWIFT: A Small-World Interaction Framework for Flow-Aware Trajectory Prediction in Autonomous Driving

T0 review · 2 major / 6 minor · reviewed 2026-07-14 · grok-4.5

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

Pith's one-line read Trajectory prediction improves when agent interactions are built from small-world network structure and traffic-flow regimes instead of pure data-driven proximity.

desk verdict Solid engineering paper: classical small-world + three-phase priors turned into a hybrid graph that actually moves the numbers on three public benchmarks. read the letter →

arxiv 2607.09741 v1 pith:SI47LWM6 submitted 2026-07-03 cs.RO cs.AIcs.MA

Chengyue Wang , Bin Rao , Haicheng Liao , Bonan Wang , Chengzhong Xu , Zhenning Li This is my paper
classification cs.ROcs.AIcs.MA
keywords trajectorypredictionautonomousdrivingsmall-worldnetworkstrafficflowtheoryinteractionmodelinggraphneuralmulti-agentforecasting
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

Most trajectory predictors for autonomous driving learn who influences whom purely from data, using proximity or attention, and therefore struggle when traffic density, location, or noise changes. This paper argues that traffic already has known structure: it behaves like a small-world network (tight local clusters plus short global paths) and its interaction patterns switch with macroscopic flow regimes (free, synchronized, congested). SWIFT hard-wires those priors. It builds an interaction graph that balances local clustering against global reach, modulates the graph parameters from scene-level density and speed statistics, and reasons over three explicit relation types (direct, co-influenced, co-influencing). On three real datasets the resulting model is more accurate, more sample-efficient, more noise-robust, and better at transferring to unseen freeways than strong baselines. The claim is that structure-aware inductive bias, not just more data or larger networks, is what produces reliable multi-agent forecasts.

What carries the argument

The Small-World Interaction Network: a hybrid optimization-plus-learning procedure that constructs a directed interaction graph by minimizing a connection-cost plus isolation-cost objective (Eq. 3), then trains a differentiable edge predictor to inherit those small-world edges; the graph parameters are further modulated by a Flow Regime Encoder derived from Optimal-Velocity-Model stability conditions.

What would settle it

Train and test on the same NGSIM split after replacing the small-world optimization target with ordinary k-nearest-neighbor edges (or after randomly shuffling the regime labels); if the accuracy, noise-robustness and cross-location gains disappear, the structural priors are not doing the claimed work.

Watch

Extended reading notes

Core claim

Embedding small-world topology and traffic-flow regimes as explicit inductive biases yields a trajectory predictor that systematically outperforms purely data-driven interaction models on accuracy, cross-location generalization, noise robustness, and sample efficiency across highway, urban, and campus scenes.

Load-bearing premise

That the hand-crafted cost minimized by simulated annealing produces a trustworthy ground-truth interaction graph that can supervise the learned edges, and that the three OVM-derived flow regimes can be labeled accurately enough from scene statistics to train the regime encoder.

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.

Referee Report

2 major / 6 minor

Summary. The paper proposes SWIFT, a trajectory-prediction framework that injects structural inductive biases from small-world network theory and three-phase traffic flow theory into interaction modeling. It constructs agent interaction graphs via a hybrid Simulated-Annealing / learned Small-World Interaction Network (Eqs. 2–12, Alg. 1), modulates graph parameters (α, β, γ) with a Flow Regime Encoder grounded in an OVM linear-stability analysis (Theorem 2, Definitions 1–3), and aggregates direct, co-influenced, and co-influencing relations through multi-channel graph convolutions (Defs. 4–6, Eqs. 76–83). Multimodal trajectories are decoded with static map, dynamic interaction, and hybrid attention/Mamba stages. On nuScenes, MoCAD, and NGSIM the model reports consistent gains over strong baselines in accuracy, plus improved sample efficiency (60% data), noise robustness, and cross-location transfer (I-80 ↔ US-101), supported by component ablations and qualitative case studies.

Significance. If the reported gains hold under independent reimplementation, the work is a solid contribution to structure-aware trajectory prediction: it couples a network-science prior (small-world connectivity) with a traffic-theoretic regime model, rather than relying solely on proximity or attention. Strengths that should be credited include (i) explicit finite-iteration SA guarantees (Theorem 1) and a full linear-stability derivation of free/synchronized/congested regimes (Theorem 2 + Lemma 1), (ii) multi-dataset evaluation spanning urban, campus, and highway regimes with complementary metrics, and (iii) non-accuracy axes—sample efficiency (Table IV), observation noise (Table V), and cross-location transfer (Tables VI–VII)—plus ablations of FRE/SWIN, fusion, φ, and loss weights. These go beyond typical accuracy-only papers and make the structure-aware claim falsifiable. The result is of clear interest to autonomous-driving prediction and to graph-based multi-agent forecasting more broadly.

major comments (2)
  1. Section III-D and IV-A4 define the regime loss Lb as cross-entropy against a one-hot ground-truth label yreg, and Section IV-A1 reports precise regime fractions (e.g., NGSIM free-flow 90.48%). The manuscript never states the concrete labeling procedure that maps raw scene statistics (density, speed, spacing) to {f,s,c} labels, nor how the unobserved OVM sensitivity a and V'(s0) in Definitions 1–3 are estimated or thresholded in practice. Without this, FRE and Lb are not reproducible and the claimed flow-aware adaptation cannot be independently verified. Please specify the labeling rule (features, thresholds or classifier, and any calibration of a) used to produce yreg on each dataset.
  2. Tables I–III and the generalization/robustness tables report single point estimates with no multi-seed means/stds or statistical tests. Given that several absolute gains are modest (e.g., nuScenes minADE5 1.24→1.15; NGSIM 1 s where BAT remains better at 0.23 vs 0.34), multi-run variance is needed to support the claim of consistent outperformance across regimes. Please add at least seed-averaged results (mean ± std) for the main tables, or a significance test against the strongest baseline per dataset.
minor comments (6)
  1. Theorem 1 restates a standard high-probability bound on the SA stationary distribution; the |S_ε| factor is typically large and the bound is not tight. A short remark that this is a formal guarantee of the optimization scheme rather than a practical mixing-time result would set expectations correctly.
  2. Eq. (11)–(12): the roles of α, β in the learned edge construction versus the SA objective (Eq. 3) are easy to conflate. Clarify whether the FRE-predicted α, β are the same scalars used in both the SA cost and the MLP fusion, or only the latter at inference.
  3. Implementation: map-free MoCAD/NGSIM substitutes road features with interaction features (IV-A3). State this substitution more prominently in the trajectory-generation section so readers do not assume HD-map inputs on those benchmarks.
  4. Fig. 6 color scale is described as normalized minADE1 but absolute ranges and the exact normalization are not given; adding a colorbar legend with numeric range would help.
  5. Typographical: “W AKE” appears with a space in Table I and text; unify to “WAKE”. “arXiv:2607.09741v1 [cs.RO] 3 Jul 2026” in the header is a future date—confirm metadata.
  6. Related work on small-world traffic (e.g., NEST, cited as [30]) is brief; a sentence contrasting SWIFT’s hybrid SA+learning graph with purely neuromodulated hypergraph designs would sharpen novelty.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: SA-derived Eop and OVM regime labels are independent structural priors, not restatements of the trajectory objective; end-to-end gains are measured on external public benchmarks.

full rationale

The derivation chain is self-contained. Structural Graph Optimization (Alg. 1, Eq. 3) minimizes a hand-crafted connection+isolation cost J(vi) that depends only on inter-agent distances ϕ; the resulting Eop supervises Ele via binary cross-entropy Lc (Eq. 99) but is never defined from future trajectories or La. Theorem 2 and Definitions 1–3 derive free/synchronized/congested regimes from the classical OVM linear-stability criterion V'(s0)<a/2; Lb is ordinary cross-entropy to those scene-statistic labels and does not recycle trajectory error. Multi-relational matrices Mp/Mi/Mt and the hybrid decoder are ordinary GNN/attention modules. All reported gains (Tables I–VII, X) are end-to-end minADE/RMSE against public ground-truth trajectories on nuScenes/MoCAD/NGSIM, with ablations that remove FRE/SWIN still leaving residual performance. Self-citations (WAKE, BAT, HLTP, NEST, GaVa) appear only as baselines or related work and are not invoked as uniqueness theorems that force the architecture. The residual score of 1 reflects only the ordinary supervised-learning loop that any graph-regularized predictor exhibits; nothing reduces a claimed prediction to its own inputs by construction.

Assumptions & free parameters 5 free parameters · 4 assumptions · 3 invented entities

The central claim rests on classical network-science and traffic-flow axioms plus a handful of free parameters that control graph construction and loss balance. No new physical entities are postulated; the invented modules are engineering constructs whose value is measured by downstream prediction metrics.

free parameters (5)
  • loss weights λa, λb, λc = 1.0 / 0.3–0.5 / 0.3–0.5
    Set by hand to (1.0,0.3,0.3) on nuScenes and (1.0,0.5,0.5) on MoCAD/NGSIM; ablation shows sensitivity of long-horizon error.
  • connection-cost exponent φ = 3
    Chosen by ablation on NGSIM; φ=3 yields best long-horizon RMSE.
  • local/global weights α, β and threshold γ = regime-dependent intervals [0.3–0.9]
    Dynamically predicted by the Flow Regime Encoder; optimal ranges reported per regime after grid search.
  • SA temperature schedule T0, η, H
    Controls quality of the “ground-truth” graph used for supervision; values not exhaustively reported.
  • number of trajectory modes K
    Standard multi-modal hyper-parameter; affects ADE/FDE metrics.
assumptions (4)
  • domain assumption Real traffic networks exhibit small-world properties (high clustering, short path lengths).
    Invoked in §I and §II-B via citation to Ghavasieh et al. and Watts–Strogatz; underpins the entire graph-construction objective.
  • domain assumption Three-phase traffic theory (free / synchronized / congested) correctly partitions interaction regimes.
    Kerner’s theory is taken as given; Definitions 1–3 and the Flow Regime Encoder rest on it.
  • domain assumption The Optimal Velocity Model linearization yields a valid stability criterion V'(s0)<a/2.
    Theorem 2 and Lemma 1 derive the regime boundaries from the OVM; the derivation assumes the classical OVM form.
  • ad hoc to paper Simulated Annealing with geometric cooling produces a sufficiently accurate approximation of the global minimizer of J(vi).
    Theorem 1 supplies only a high-probability bound; the practical quality of the SA “ground truth” is assumed adequate for supervision.
invented entities (3)
  • Small-World Interaction Network (hybrid SA + learned edge set)
    purpose: Constructs a directed interaction graph that simultaneously enforces local clustering and short global paths.
    Engineering construct; no independent physical existence claimed beyond the prediction task.
  • Flow Regime Encoder
    purpose: Maps scene-level statistics to the three interaction parameters α, β, γ.
    Learned module whose utility is measured only by downstream ADE/RMSE; no external falsifiable prediction.
  • Co-Influenced and Co-Influencing adjacency matrices Mi, Mt
    purpose: Capture second-order relational motifs beyond pairwise edges.
    Graph-theoretic constructs introduced for message passing; evaluated only inside the prediction pipeline.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SWIFT: A Small-World Interaction Framework for Flow-Aware Trajectory Prediction in Autonomous Driving." pith.science (2026). https://pith.science/paper/SI47LWM6

@misc{pith2026260709741,
  author       = {Pith},
  title        = {Pith review of: SWIFT: A Small-World Interaction Framework for Flow-Aware Trajectory Prediction in Autonomous Driving},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SI47LWM6}},
  note         = {Machine review of arXiv:2607.09741}
}
read the original abstract

Accurate trajectory prediction in autonomous driving hinges on modeling dynamic and context-dependent interactions among traffic agents. However, most existing approaches are purely data-driven and lack structural priors, which limits their generalization under distribution shifts. In this work, interaction modeling is revisited through the structure and dynamics of traffic networks, and SWIFT (Small-World Interaction Framework for Trajectory prediction) is proposed as a unified framework that integrates small-world networks with traffic flow theory. SWIFT introduces structural inductive biases via a Small-World Interaction Network that captures both local and global dependencies, and a Flow Regime Encoder that adapts the interaction structure to scene-level traffic states. Interaction reasoning is further enhanced through a multi-relational graph module that explicitly encodes direct and higher-order agent relationships. Extensive experiments on three real-world datasets, nuScenes, MoCAD, and NGSIM, show that SWIFT consistently outperforms strong baselines in prediction accuracy across diverse traffic regimes. Beyond accuracy, SWIFT exhibits improved generalization to unseen locations and regimes, robustness under noisy observations, and strong performance with limited training data, supporting the effectiveness of its structure-aware design.

Figures

Figures reproduced from arXiv: 2607.09741 by the authors.

Figure 1
Figure 1. Conceptual comparison between (a) existing approach [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Spatial-temporal visualization and regime-specific [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Overview of the proposed SWIFT model. Panel (a) presents the overall workflow, which takes agents’ historical [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Illustration of the training process for Small-world [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Qualitative comparison of trajectory prediction and [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]
Figure 6
Figure 6. Figure 6: Grid search results of the interaction parameters [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]
Figure 7
Figure 7. Figure 7: Qualitative comparison between SWIFT and its ablated [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: Failure cases of SWIFT. These failures arise when the [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

68 extracted references · 3 linked inside Pith

  1. [1]

    Smemo: social memory for trajectory forecasting,

    F. Marchetti, F. Becattini, L. Seidenari, and A. Del Bimbo, “Smemo: social memory for trajectory forecasting,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 46, no. 6, pp. 4410–4425, 2024

    2024

  2. [2]

    Modality exploration, retrieval and adaptation for trajectory prediction,

    J. Sun, Y . Li, L. Chai, and C. Lu, “Modality exploration, retrieval and adaptation for trajectory prediction,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 45, no. 12, pp. 15 051–15 064, 2023

    2023

  3. [3]

    Decoupling motion forecasting into directional intentions and dynamic states,

    B. Zhang, N. Song, and L. Zhang, “Decoupling motion forecasting into directional intentions and dynamic states,” NIPS, 2024

    2024

  4. [4]

    End-point drive and reverse enhanced decoding-based traffic participants trajectory prediction under bird’s eye view,

    C. Liu, J. Xie, F. Chang, T. Li, S. Li, and Y . Wang, “End-point drive and reverse enhanced decoding-based traffic participants trajectory prediction under bird’s eye view,”IEEE Transactions on Intelligent Transportation Systems, 2025

    2025

  5. [5]

    Sa-tp: A safety-aware trajectory prediction and planning model for autonomous driving,

    H. Liao, Z. Li, K. Zhu, K. Li, and C. Xu, “Sa-tp: A safety-aware trajectory prediction and planning model for autonomous driving,” IEEE Transactions on Robotics , 2025

    2025

  6. [6]

    Social force embedded mixed graph convolutional network for multi-class trajectory prediction,

    Q. Du, X. Wang, S. Yin, L. Li, and H. Ning, “Social force embedded mixed graph convolutional network for multi-class trajectory prediction,” IEEE Transactions on Intelligent Vehicles , 2024

    2024

  7. [7]

    Heterogeneous graph social pooling for interaction-aware vehicle trajectory prediction,

    X. Mo, Y . Xing, and C. Lv, “Heterogeneous graph social pooling for interaction-aware vehicle trajectory prediction,”Transportation Research Part E: Logistics and Transportation Review, vol. 191, p. 103748, 2024

    2024

  8. [8]

    Diversity of information pathways drives sparsity in real-world networks,

    A. Ghavasieh and M. De Domenico, “Diversity of information pathways drives sparsity in real-world networks,” Nature Physics, vol. 20, no. 3, pp. 512–519, 2024

    2024

Show all 68 references
  1. [9]

    B. S. Kerner, Introduction to modern traffic flow theory and control: the long road to three-phase traffic theory . Springer Science & Business Media, 2009

    2009

  2. [10]

    Three-phase traffic theory and highway capacity,

    ——, “Three-phase traffic theory and highway capacity,” Physica A: Statistical Mechanics and its Applications , vol. 333, pp. 379–440, 2004

    2004

  3. [11]

    Attention-aware social graph transformer networks for stochastic trajectory prediction,

    Y . Liu, B. Li, X. Wang, C. Sammut, and L. Yao, “Attention-aware social graph transformer networks for stochastic trajectory prediction,” IEEE Transactions on Knowledge and Data Engineering , 2024

    2024

  4. [12]

    Intention- aware vehicle trajectory prediction based on spatial-temporal dynamic attention network for internet of vehicles,

    X. Chen, H. Zhang, F. Zhao, Y . Hu, C. Tan, and J. Yang, “Intention- aware vehicle trajectory prediction based on spatial-temporal dynamic attention network for internet of vehicles,” IEEE Transactions on Intel- ligent Transportation Systems, vol. 23, no. 10, pp. 19 471–19 483, 2022

    2022

  5. [13]

    Helbing, Social self-organization: Agent-based simulations and ex- periments to study emergent social behavior

    D. Helbing, Social self-organization: Agent-based simulations and ex- periments to study emergent social behavior . Springer, 2012

    2012

  6. [14]

    Dstigcn: Deformable spatial- temporal interaction graph convolution network for pedestrian trajectory prediction,

    W. Chen, H. Sang, J. Wang, and Z. Zhao, “Dstigcn: Deformable spatial- temporal interaction graph convolution network for pedestrian trajectory prediction,” IEEE Transactions on Intelligent Transportation Systems , 2025

    2025

  7. [15]

    Mtp-go: Graph- based probabilistic multi-agent trajectory prediction with neural odes,

    T. Westny, J. Oskarsson, B. Olofsson, and E. Frisk, “Mtp-go: Graph- based probabilistic multi-agent trajectory prediction with neural odes,” IEEE Transactions on Intelligent Vehicles , 2023

    2023

  8. [16]

    Us highway 101 dataset,

    G. SIMulation, “Us highway 101 dataset,” 2007

    2007

  9. [17]

    Hi-scl: Fighting long-tailed challenges in trajectory prediction with hierarchical wave-semantic contrastive learning,

    Z. Lan, Y . Ren, H. Yu, L. Liu, Z. Li, Y . Wang, and Z. Cui, “Hi-scl: Fighting long-tailed challenges in trajectory prediction with hierarchical wave-semantic contrastive learning,” Transportation Research Part C: Emerging Technologies, vol. 165, p. 104735, 2024

    2024

  10. [18]

    Physics- informed trajectory prediction for autonomous driving under missing observation,

    H. Liao, C. Wang, Z. Li, Y . Li, B. Wang, G. Li, and C. Xu, “Physics- informed trajectory prediction for autonomous driving under missing observation,” International Joint Conference On Artificial Intelligence , 2024

    2024

  11. [19]

    Social lstm: Human trajectory prediction in crowded spaces,

    A. Alahi, K. Goel, V . Ramanathan, A. Robicquet, L. Fei-Fei, and S. Savarese, “Social lstm: Human trajectory prediction in crowded spaces,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 961–971

    2016

  12. [20]

    A cognitive-based trajectory prediction approach for autonomous driving,

    H. Liao, Y . Li, Z. Li, C. Wang, Z. Cui, S. E. Li, and C. Xu, “A cognitive-based trajectory prediction approach for autonomous driving,” IEEE Transactions on Intelligent Vehicles , 2024

    2024

  13. [21]

    Multimodal trajectory prediction conditioned on lane-graph traversals,

    N. Deo, E. Wolff, and O. Beijbom, “Multimodal trajectory prediction conditioned on lane-graph traversals,” in Conference on Robot Learning. PMLR, 2022, pp. 203–212

    2022

  14. [22]

    Q-eanet: Implicit social modeling for trajectory prediction via experience-anchored queries,

    J. Chen, Z. Wang, J. Wang, and B. Cai, “Q-eanet: Implicit social modeling for trajectory prediction via experience-anchored queries,” IET Intelligent Transport Systems, vol. 18, no. 6, pp. 1004–1015, 2024

    2024

  15. [23]

    Toward human-like trajec- tory prediction for autonomous driving: A behavior-centric approach,

    H. Liao, Z. Li, G. Zhang, K. Li, and C. Xu, “Toward human-like trajec- tory prediction for autonomous driving: A behavior-centric approach,” Transportation Science, 2025

    2025

  16. [24]

    Nettraj: A network-based vehicle trajectory prediction model with directional representation and spatiotemporal attention mechanisms,

    Y . Liang and Z. Zhao, “Nettraj: A network-based vehicle trajectory prediction model with directional representation and spatiotemporal attention mechanisms,” IEEE Transactions on Intelligent Transportation Systems, vol. 23, no. 9, pp. 14 470–14 481, 2021

    2021

  17. [25]

    Collective dynamics of ‘small- world’networks,

    D. J. Watts and S. H. Strogatz, “Collective dynamics of ‘small- world’networks,”nature, vol. 393, no. 6684, pp. 440–442, 1998

    1998

  18. [26]

    Cultural and opinion dynamics in small-world “social

    Y . Lee, “Cultural and opinion dynamics in small-world “social” net- works,” The Journal of Mathematical Sociology , pp. 1–26, 2024

    2024

  19. [27]

    Navigation in a small world,

    J. M. Kleinberg, “Navigation in a small world,” Nature, vol. 406, no. 6798, pp. 845–845, 2000

    2000

  20. [28]

    Electric vehicle charging planning: A complex systems perspective,

    A. P. Zhao, S. Li, Z. Li, Z. Wang, X. Fei, Z. Hu, M. Alhazmi, X. Yan, C. Wu, S. Lu et al. , “Electric vehicle charging planning: A complex systems perspective,” IEEE Transactions on Smart Grid , 2024

    2024

  21. [29]

    A shockwave profile model for traffic flow on congested urban arterials,

    X. Wu and H. X. Liu, “A shockwave profile model for traffic flow on congested urban arterials,” Transportation Research Part B: Method- ological, vol. 45, no. 10, pp. 1768–1786, 2011

    2011

  22. [30]

    Nest: A neuromodulated small-world hypergraph trajectory prediction model for autonomous driving,

    C. Wang, H. Liao, B. Wang, Y . Guan, B. Rao, Z. Pu, Z. Cui, C.-Z. Xu, and Z. Li, “Nest: A neuromodulated small-world hypergraph trajectory prediction model for autonomous driving,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 39, no. 1, 2025, pp. 808–816

    2025

  23. [31]

    Bat: Behavior-aware human-like trajectory prediction for autonomous driv- ing,

    H. Liao, Z. Li, H. Shen, W. Zeng, D. Liao, G. Li, and C. Xu, “Bat: Behavior-aware human-like trajectory prediction for autonomous driv- ing,” in Proceedings of the AAAI Conference on Artificial Intelligence , vol. 38, 2024, pp. 10 332–10 340

    2024

  24. [32]

    Hivt: Hierarchical vector transformer for multi-agent motion prediction,

    Z. Zhou, L. Ye, J. Wang, K. Wu, and K. Lu, “Hivt: Hierarchical vector transformer for multi-agent motion prediction,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 8823–8833

    2022

  25. [33]

    Small-world formation via local information,

    S. Alamdari, “Small-world formation via local information,” arXiv preprint arXiv:2301.00849, 2023

    2023 arXiv

  26. [34]

    The physics of traffic,

    B. S. Kerner, “The physics of traffic,” Physics world , vol. 12, no. 8, p. 25, 1999

    1999

  27. [35]

    Dynamical model of traffic congestion and numerical simulation,

    M. Bando, K. Hasebe, A. Nakayama, A. Shibata, and Y . Sugiyama, “Dynamical model of traffic congestion and numerical simulation,” Physical review E , vol. 51, no. 2, p. 1035, 1995

    1995

  28. [36]

    Analysis of optimal velocity model with explicit delay,

    M. Bando, K. Hasebe, K. Nakanishi, and A. Nakayama, “Analysis of optimal velocity model with explicit delay,” Physical Review E, vol. 58, no. 5, p. 5429, 1998

    1998

  29. [37]

    Speed or spacing? cumulative variables, and convolution of model errors and time in traffic flow models valida- tion and calibration,

    V . Punzo and M. Montanino, “Speed or spacing? cumulative variables, and convolution of model errors and time in traffic flow models valida- tion and calibration,” Transportation Research Part B: Methodological , vol. 91, pp. 21–33, 2016

    2016

  30. [38]

    Fourier transforms and the fast fourier transform (fft) algorithm,

    P. Heckbert, “Fourier transforms and the fast fourier transform (fft) algorithm,” Computer Graphics, vol. 2, no. 1995, pp. 15–463, 1995

    1995

  31. [39]

    Car-following models: fifty years of linear stability analysis–a mathematical perspective,

    R. E. Wilson and J. A. Ward, “Car-following models: fifty years of linear stability analysis–a mathematical perspective,” Transportation Planning and Technology, vol. 34, no. 1, pp. 3–18, 2011

    2011

  32. [40]

    Mechanisms for spatio-temporal pattern formation in highway traffic models,

    R. E. Wilson, “Mechanisms for spatio-temporal pattern formation in highway traffic models,”Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, vol. 366, no. 1872, pp. 2017–2032, 2008

    2017

  33. [41]

    Coherent moving states in highway traffic,

    D. Helbing and B. A. Huberman, “Coherent moving states in highway traffic,” Nature, vol. 396, no. 6713, pp. 738–740, 1998

    1998

  34. [42]

    An extension of parseval’s theorem and its use in calculating transient energy in the frequency domain,

    S. Kelkar, L. Grigsby, and J. Langsner, “An extension of parseval’s theorem and its use in calculating transient energy in the frequency domain,” IEEE Transactions on Industrial Electronics , no. 1, pp. 42– 45, 2007

    2007

  35. [43]

    Directed graph convolutional network,

    Z. Tong, Y . Liang, C. Sun, D. S. Rosenblum, and A. Lim, “Directed graph convolutional network,” arXiv preprint arXiv:2004.13970 , 2020

    2004 arXiv

  36. [44]

    nuscenes: A multimodal dataset for autonomous driving,

    H. Caesar, V . Bankiti, A. H. Lang, S. V ora, V . E. Liong, Q. Xu, A. Krishnan, Y . Pan, G. Baldan, and O. Beijbom, “nuscenes: A multimodal dataset for autonomous driving,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 11 621–11 631

    2020

  37. [45]

    Ngsim—next generation simulation,

    U. D. of Transportation, “Ngsim—next generation simulation,” 2008

    2008

  38. [46]

    Wake: Towards robust and physically feasible trajectory prediction for autonomous vehicles with wavelet and kinematics synergy,

    C. Wang, H. Liao, Z. Li, and C. Xu, “Wake: Towards robust and physically feasible trajectory prediction for autonomous vehicles with wavelet and kinematics synergy,”IEEE Transactions on Pattern Analysis and Machine Intelligence , 2025

    2025

  39. [47]

    Trajectron++: Dynamically-feasible trajectory forecasting with heterogeneous data,

    T. Salzmann, B. Ivanovic, P. Chakravarty, and M. Pavone, “Trajectron++: Dynamically-feasible trajectory forecasting with heterogeneous data,” in Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XVIII 16 . Springer, 2020, p...

    2020

  40. [48]

    Multipath: Multiple probabilistic anchor trajectory hypotheses for behavior prediction,

    Y . Chai, B. Sapp, M. Bansal, and D. Anguelov, “Multipath: Multiple probabilistic anchor trajectory hypotheses for behavior prediction,” in Conference on Robot Learning . PMLR, 2020, pp. 86–99

    2020

  41. [49]

    Agentformer: Agent-aware transformers for socio-temporal multi-agent forecasting,

    Y . Yuan, X. Weng, Y . Ou, and K. Kitani, “Agentformer: Agent-aware transformers for socio-temporal multi-agent forecasting,” 2021. JOURNAL OF IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, JANUARY 2026 16

    2021

  42. [50]

    Lapred: Lane-aware prediction of multi-modal future trajectories of dynamic agents,

    B. Kim, S. H. Park, S. Lee, E. Khoshimjonov, D. Kum, J. Kim, J. S. Kim, and J. W. Choi, “Lapred: Lane-aware prediction of multi-modal future trajectories of dynamic agents,” in CVPR, 2021, pp. 14 636–14 645

    2021

  43. [51]

    Stgm: Vehicle trajectory prediction based on generative model for spatial-temporal features,

    Z. Zhong, Y . Luo, and W. Liang, “Stgm: Vehicle trajectory prediction based on generative model for spatial-temporal features,” IEEE Transac- tions on Intelligent Transportation Systems , vol. 23, no. 10, pp. 18 785– 18 793, 2022

    2022

  44. [52]

    Gohome: Graph-oriented heatmap output for future motion estimation,

    T. Gilles, S. Sabatini, D. Tsishkou, B. Stanciulescu, and F. Moutarde, “Gohome: Graph-oriented heatmap output for future motion estimation,” in ICRA. IEEE, 2022, pp. 9107–9114

    2022

  45. [53]

    Context-aware timewise vaes for real-time vehicle trajectory prediction,

    P. Xu, J.-B. Hayet, and I. Karamouzas, “Context-aware timewise vaes for real-time vehicle trajectory prediction,” IEEE Robotics and Automation Letters, 2023

    2023

  46. [54]

    Emsin: Enhanced multi-stream interaction network for vehicle trajectory prediction,

    Y . Ren, Z. Lan, L. Liu, and H. Yu, “Emsin: Enhanced multi-stream interaction network for vehicle trajectory prediction,” IEEE Transactions on Fuzzy Systems , 2024

    2024

  47. [55]

    Cot-drive: Efficient motion forecasting for autonomous driving with llms and chain-of-thought prompting,

    H. Liao, H. Kong, B. Wang, C. Wang, W. Ye, Z. He, C. Xu, and Z. Li, “Cot-drive: Efficient motion forecasting for autonomous driving with llms and chain-of-thought prompting,” IEEE Transactions on Artificial Intelligence, 2025

    2025

  48. [56]

    Seflow: A self-supervised scene flow method in autonomous driving,

    Q. Zhang, Y . Yang, P. Li, O. Andersson, and P. Jensfelt, “Seflow: A self-supervised scene flow method in autonomous driving,” in European Conference on Computer Vision . Springer, 2024, pp. 353–369

    2024

  49. [57]

    Social gan: Socially acceptable trajectories with generative adversarial networks,

    A. Gupta, J. Johnson, L. Fei-Fei, S. Savarese, and A. Alahi, “Social gan: Socially acceptable trajectories with generative adversarial networks,” in CVPR, 2018, pp. 2255–2264

    2018

  50. [58]

    Convolutional social pooling for vehicle tra- jectory prediction,

    N. Deo and M. M. Trivedi, “Convolutional social pooling for vehicle tra- jectory prediction,” in Proceedings of the IEEE conference on computer vision and pattern recognition workshops , 2018, pp. 1468–1476

    2018

  51. [59]

    Non-local social pooling for vehicle trajectory prediction,

    K. Messaoud, I. Yahiaoui, A. Verroust-Blondet, and F. Nashashibi, “Non-local social pooling for vehicle trajectory prediction,” in 2019 IEEE Intelligent Vehicles Symposium (IV) . IEEE, 2019, pp. 975–980

    2019

  52. [60]

    Attention based vehicle trajectory prediction,

    ——, “Attention based vehicle trajectory prediction,” IEEE Transactions on Intelligent Vehicles, vol. 6, no. 1, pp. 175–185, 2021

    2021

  53. [61]

    Congestion-aware multi-agent trajectory prediction for collision avoidance,

    X. Xie, C. Zhang, Y . Zhu, Y . N. Wu, and S.-C. Zhu, “Congestion-aware multi-agent trajectory prediction for collision avoidance,” inICRA, 2021

    2021

  54. [62]

    Wsip: wave superposition inspired pooling for dynamic interactions-aware trajectory prediction,

    R. Wang, S. Wang, H. Yan, and X. Wang, “Wsip: wave superposition inspired pooling for dynamic interactions-aware trajectory prediction,” in AAAI, vol. 37, 2023, pp. 4685–4692

    2023

  55. [63]

    Multi-agent tensor fusion for contextual trajectory prediction,

    T. Zhao, Y . Xu, M. Monfort, W. Choi, C. Baker, Y . Zhao, Y . Wang, and Y . N. Wu, “Multi-agent tensor fusion for contextual trajectory prediction,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2019, pp. 12 126–12 134

    2019

  56. [64]

    Vehicle trajec- tory prediction based on intention-aware non-autoregressive transformer with multi-attention learning for internet of vehicles,

    X. Chen, H. Zhang, F. Zhao, Y . Cai, H. Wang, and Q. Ye, “Vehicle trajec- tory prediction based on intention-aware non-autoregressive transformer with multi-attention learning for internet of vehicles,”IEEE Transactions on Instrumentation and Measurement , vol. 71, pp. 1–12, 2022

    2022

  57. [65]

    Trajectory prediction network of autonomous vehicles with fusion of historical interactive features,

    Z. Zuo, X. Wang, S. Guo, Z. Liu, Z. Li, and Y . Wang, “Trajectory prediction network of autonomous vehicles with fusion of historical interactive features,” IEEE Transactions on Intelligent Vehicles , 2023

    2023

  58. [66]

    Dacr-amtp: Adaptive multi-modal vehicle trajectory prediction for dynamic drivable areas based on collision risk,

    P. Cong, Y . Xiao, X. Wan, M. Deng, J. Li, and X. Zhang, “Dacr-amtp: Adaptive multi-modal vehicle trajectory prediction for dynamic drivable areas based on collision risk,” IEEE Transactions on Intelligent Vehicles, 2023

    2023

  59. [67]

    Human observation-inspired trajectory prediction for autonomous driving in mixed-autonomy traffic environments,

    H. Liao, S. Liu, Y . Li, Z. Li, C. Wang, Y . Li, S. E. Li, and C. Xu, “Human observation-inspired trajectory prediction for autonomous driving in mixed-autonomy traffic environments,” in ICRA. IEEE, 2024, pp. 14 212–14 219

    2024

  60. [68]

    Estimating or propagating gradients through stochastic neurons for conditional computation,

    Y . Bengio, N. L ´eonard, and A. Courville, “Estimating or propagating gradients through stochastic neurons for conditional computation,” arXiv preprint arXiv:1308.3432, 2013. Chengyue Wang is currently pursuing a Ph.D. degree at the State Key Laboratory of Internet of Things ...

    2013 arXiv

Pith tools

Reviewed July 14, 2026 · model on record in the stance chip above.