Pith. sign in

REVIEW 5 major objections 5 minor 2 cited by

InterHub: A Naturalistic Trajectory Dataset with Dense Interaction for Autonomous Driving

T0 review · 5 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A formal definition of driving interaction mines 65,399 dense events from four public datasets and shows these events stress-test current predictors and planners.

desk verdict A potentially useful dense-interaction dataset and toolkit, but the extraction rule is unvalidated, the benchmark is partly circular, and Table 2 contains an impossible miss rate. read the letter →

arxiv 2411.18302 v2 pith:Y64P57Z3 submitted 2024-11-27 cs.RO

Xiyan Jiang , Xiaocong Zhao , Yiru Liu , Zirui Li , Peng Hang , Lu Xiong , Jian Sun This is my paper · ORCID
classification cs.RO
keywords drivinginteractiontrajectorydatasetnaturalisticdataMetricTemporalLogicintensitymotionpredictionplanningformalmethods
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

InterHub claims to solve a data shortage: genuine driving-interaction events are so sparse in naturalistic records that interaction-focused research lacks a large, unified source of examples. The paper proposes a formal definition of driving interaction in Metric Temporal Logic, with the minimum sum of absolute accelerations required to resolve predicted conflicts as a quantitative measure of whether an interaction happens and how intense it is. Applied to Waymo, nuPlan, Lyft, and INTERACTION records, this definition yields 65,399 interaction events involving 146,170 vehicle participants, unified into a single indexed dataset. The paper further claims that these interaction-dense segments are precisely where current trajectory predictors and motion planners degrade, and it releases a toolkit for extracting the same kind of events from other public or private data.

What carries the argument

The load-bearing quantity is $\mathrm{MSAA}_t$, the minimum sum of absolute accelerations that all potentially involved vehicles must take to resolve their predicted conflicts, computed by optimizing over each vehicle's required longitudinal acceleration subject to state-transition, velocity, and time-interval constraints. The extraction rule is encoded in Metric Temporal Logic (MTL): constant-speed five-second future trajectories, a three-second conflict-time window, buffer-polygon intersection tests, chain-conflict propagation across agent pairs, and the condition that $\mathrm{MSAA}_t$ exceeds a threshold together fix the participating agents and the boundaries of each interaction segment. The same $\mathrm{MSAA}_t$ value doubles as the interaction-intensity label attached to every event.

What would settle it

Take a random sample of extracted events and an equal number of non-extracted clips from the same datasets, have independent human raters mark whether vehicles are mutually influencing one another, and compare their labels with the formal rule's predictions. The claim fails if high-effort events routinely involve vehicles that simply continue at constant speed without reacting, or if events that raters clearly call interactions are missed because the five-second projection or three-second window excludes them.

Watch

Extended reading notes

Core claim

The central discovery is a formal, quantitative criterion for what counts as a driving interaction, together with a dataset built from it. Two agents are in spatiotemporal conflict if their constant-speed future trajectories over the next five seconds intersect and their arrival times at the intersection differ by less than three seconds; agents connected through chains of such conflicts form the potentially involved set. At each time step, the paper solves an optimization for the minimum sum of absolute accelerations ($\mathrm{MSAA}_t$) the involved agents must apply to pass through their conflict points, and an interaction exists when this minimal effort exceeds a threshold. Metric Temporal Logic formulas wrap this condition into complete interaction segments by fixing the involved agents and the start and end times. The extraction yields 65,399 events from four naturalistic datasets, with mean intensity 1.76 m/s² and mean duration 1.13 s, and the interaction-only test sets push current predictors and planners into substantial performance degradation, with miss rate, minADE, and minFDE rising across all evaluated baselines.

Load-bearing premise

The extraction rule—constant-speed five-second future paths, a three-second conflict-time window, the buffer-based intersection test, and the condition that minimal conflict-resolution effort exceeds a threshold—is assumed to capture exactly the situations that count as driving interactions, but it is not validated against human-labeled interaction boundaries or an independent interaction measure.

Editorial extensions

If this is right

  • A single unified dataset of 65,399 interaction events from four public sources becomes available, so interaction studies can be compared across datasets instead of resting on one source.
  • The formal MTL definition and open toolkit let researchers expand the dataset with new public or private trajectory data, making the benchmark extendable.
  • Interaction-only test sets expose weaknesses hidden in full-dataset averages: UQnet's miss rate rises from 0.60 to 1.65, MTR's from 0.19 to 0.36, and ContextVAE's minADE more than doubles on Lyft.
  • Planning performance falls as interaction intensity rises: top-ranked planners in a 700-scenario challenge score lower and fail to complete the task more often on high-intensity events.
  • Because only about 3% of scenarios contain two-agent interactions and about 1% contain multi-agent interactions, dense interaction segments are the informative tail of naturalistic driving data for safety-critical validation.

Reading between the lines

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

  • If the formal extraction is accepted, $\mathrm{MSAA}_t$ could serve as a continuous, scenario-agnostic difficulty score for autonomous-vehicle testing, complementing safety surrogates such as PET and time-to-collision; this extension is implied by the planning results but not developed in the paper.
  • The framework is demonstrated on motor vehicles, but the underlying definition of interaction extends to pedestrians and cyclists; testing whether the MSAA conflict machinery captures interactions involving vulnerable road users would be a natural next experiment.
  • The sharp performance drop on interaction-only segments suggests that full-dataset leaderboard rankings may overstate deployment readiness; a standardized interaction-only test set could become a routine stress test for prediction and planning systems.
  • Because the toolkit is open, an obvious check is to mine interaction events from other public trajectory collections and see whether the same intensity distributions and the same predictor-degradation pattern reappear.
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

5 major / 5 minor

Summary. The paper introduces InterHub, a dataset of 65,399 driving-interaction events extracted from Waymo, nuPlan, Lyft, and INTERACTION. The extraction is based on a formal-methods-inspired definition of interaction: agents are linked through spatiotemporal conflict points, and interaction intensity is measured by the minimum sum of absolute accelerations (MSAA) needed to resolve the projected conflicts. The paper also provides a data-unification toolkit and reports experiments showing that trajectory predictors and planners perform worse on the 1000 highest-intensity interaction segments than on the full test sets. The authors argue that InterHub supplies a dense, unified, and reusable resource for interaction-focused autonomous driving research.

Significance. If the extraction method were fully specified and validated, InterHub would be a valuable community resource: it unifies four major driving datasets into a common format, provides an open toolkit, and offers a large set of interaction segments with intensity labels, PET values, and participant counts. The analysis of interaction sparsity and the planning-challenge correlation are useful observations. However, the central extraction rule is not reproducible from the paper as written, the formal definition contains ambiguous or vacuous constraints, no independent validation of the extracted events is provided, and the benchmark table contains a metric error and an impossible MR value. These issues are load-bearing because they affect the validity of the dataset itself and the benchmark conclusions.

major comments (5)
  1. [Extracting interaction events using the formal method] The activation threshold for IntCheck is never specified. In the 'Complete interactive driving segment' formula, IntCheck is defined as 'MSAA > threshold', while Figure 5 states that MSAAt is 'non-zero'; no numerical threshold is given anywhere. Since every interaction event in the dataset is determined by this threshold, the extraction is not reproducible from the paper alone. Please report the exact threshold used, reconcile the 'non-zero' and '> threshold' formulations, and provide a sensitivity analysis over the threshold, the projection horizon m, the conflict-time window ConfTime, and the buffer size n used in Intersection.
  2. [Extracting interaction events using the formal method, Figure 5] The Time Interval Constraint as stated is vacuous: for any two real-valued arrival times, the condition that their time interval is 'non-negative' is always satisfied. A meaningful conflict-resolution constraint would be an upper bound on the absolute time difference, or a required safety margin. Relatedly, the optimization problem that defines MSAAt is not written out; the text only names constraints. Without the explicit objective and constraints, MSAA is not formally defined and the reported intensity values cannot be independently checked. Please state the full optimization program, including the exact meaning of every variable in Figure 5.
  3. [Technical validation] The extracted events are never validated against human-annotated interaction boundaries or an independent interaction measure such as the MTTCP used in the INTERACTION dataset. The paper's central claim is that the 65,399 events are genuine driving interactions, but the only evidence offered is the internal consistency of the extraction rule and a PET analysis that is not a precision/recall test. Please add a validation study, for example by comparing extracted segments against human labels on a sample, or by measuring agreement with an independent interaction detector.
  4. [Technical validation, Table 2 and Eq. (3)] The Miss Rate definition in Eq. (3) is inconsistent with its name: the indicator returns 1 when the minimum final distance is smaller than the threshold δ, which counts successful predictions, not misses. In addition, Table 2 reports MR = 1.65 for UQnet on InterHub, which is impossible for any rate defined as a proportion in [0,1]. These errors undermine the quantitative benchmark conclusions; please correct the metric definition and rerun or re-report the benchmark results.
  5. [Versatile benchmark for key autonomous driving tasks] The benchmark selects the 1000 segments with the highest MSAA intensity using the same metric that defines the interaction events. Observing performance degradation on these segments is partly a consequence of the selection rule, and it does not independently demonstrate that InterHub contains especially challenging interactions. Please temper the claim or compare against a baseline selected by an independent criterion such as PET, TTC, or human-annotated difficulty.
minor comments (5)
  1. [Background & Summary, Figure 1] The word 'pedestrain' in Figure 1 is a typo and should be 'pedestrian'.
  2. [Motion prediction, UQnet description] The phrase 'significantly enhancing the generalizability of the missing rate' is unclear; presumably it should refer to 'miss rate' or a related metric.
  3. [Table 2] The claim of significance 'at the confidence level of α = 0.05' is not supported by any description of the statistical test used or the confidence intervals; please provide details or remove the claim.
  4. [Data records] The figshare repository is mentioned but no URL is given; please include the persistent identifier for the dataset.
  5. [Figure 5] The text in Figure 5 is garbled in places, such as 'the distance vehicle i travels at t' and 'PotConfPoint(3)'; the notation should be made self-contained and consistent with the main text.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular steps found; the extraction rule is a stipulative formalization, and the benchmark uses InterHub as an external testbed rather than as a fitted input.

full rationale

The paper's central artifact, InterHub, is produced by a stated formal definition: interactions are events where MSAA > threshold, with MTL conditions for spatiotemporal conflict and min-effort conflict resolution. This is a stipulative definition, not a derivation of one quantity from another that already contains the target result. The paper does not claim to predict an independent interaction measure from fitted parameters, so the self-definitional and fitted-input patterns do not apply. The benchmark selects the 1000 highest-intensity InterHub segments and reports performance degradation of trained predictors and planners; this is an empirical evaluation on a deliberately stressed subset, not a logical consequence of the selection rule. The models could in principle maintain performance, and the observed degradation is a real generalization result. The citation of UQnet as 'our previous work' is only a baseline-model choice, not an unverified load-bearing premise, and no uniqueness theorem or ansatz is imported from the authors' prior work. Concerns about the unspecified MSAA threshold, the ambiguous Time Interval Constraint, the lack of human-annotation validation, and the impossible MR value of 1.65 in Table 2 are reproducibility, validity, or correctness issues, but they are not instances of circularity under the required definition of a step reducing to its own inputs. Therefore no significant circularity is present; the appropriate score is 0.

Assumptions & free parameters 5 free parameters · 7 assumptions · 1 invented entities

The dataset and benchmark rest on several hand-picked thresholds (5-second horizon, 3-second conflict window, trajectory buffer size, MSAA activation threshold, 1000-segment test subset) and on a chain of modeling assumptions: Markkula's definition, constant-speed path projection, longitudinal-only conflict resolution, and the validity of the four source datasets. None of the thresholds are justified by sensitivity analyses, and no external ground truth is used to validate the extraction. The MSAA metric is a new construct without independent calibration.

free parameters (5)
  • Trajectory projection horizon m = 5 s
    Used in FuTraj(obj_i, m) to estimate future paths; chosen by hand and no sensitivity analysis is reported.
  • Conflict time threshold ConfTime = 3 s
    Defines temporal proximity at conflict points in TimeDiffLimit; hand-set threshold not varied or justified.
  • Trajectory buffer size n = not reported
    Used in BufferLine and BufferPolygon for Intersection; value is not stated in the text.
  • MSAA activation threshold = not reported
    IntCheck requires MSAAt > threshold; the actual threshold used to count interaction events is not given.
  • Benchmark subset size = 1000
    Top 1000 highest-intensity segments form the interaction-only test set; this arbitrary size affects all reported degradation numbers.
assumptions (7)
  • domain assumption Markkula et al.'s qualitative definition of driving interaction is accepted as the ground truth.
    The formal method is built on the Markkula definition cited as ref. 31; if this definition is not a valid basis, the extraction targets are misaligned.
  • domain assumption Future trajectories can be approximated by constant-speed motion over a fixed 5-second horizon.
    Assumed in FuTraj when computing conflict points; in reality drivers brake, accelerate, and change lanes reactively.
  • domain assumption Interaction intensity equals the minimum sum of absolute longitudinal accelerations (MSAA) required to resolve conflicts.
    This is the core quantitative proxy; no evidence is provided that MSAA correlates with human judgments or objective safety beyond PET plots.
  • domain assumption All involved agents can adjust via longitudinal acceleration only.
    The optimization variables are longitudinal accelerations; lateral maneuvers and strategic choices are not modeled, which may misrepresent real resolutions.
  • domain assumption The four source datasets provide sufficiently accurate trajectories and maps for interaction mining.
    Raw data quality is taken on faith from Waymo, nuPlan, Lyft, and INTERACTION and unified through trajdata without independent verification.
  • standard math Metric Temporal Logic operators are used with standard semantics.
    The paper uses G and F operators with intervals; this is standard, but the encoding is not formalized in a proof assistant.
  • ad hoc to paper The MSAA optimization problem has a computable, unique minimum for every scenario.
    No solver details, convexity proof, or feasibility discussion is provided, yet the extraction depends on solving this problem for every timestep.
invented entities (1)
  • MSAA (minimum sum of absolute accelerations) intensity metric
    purpose: Quantifies interaction intensity as the minimal collective longitudinal acceleration effort needed to resolve spatial-temporal conflicts.
    Introduced in this paper as a new scalar metric; it is not calibrated against independent human interaction ratings or an established intensity measure.

how reviews work

0 comments
Cite this review

Pith. "Pith review of InterHub: A Naturalistic Trajectory Dataset with Dense Interaction for Autonomous Driving." pith.science (2026). https://pith.science/paper/Y64P57Z3

@misc{pith2026241118302,
  author       = {Pith},
  title        = {Pith review of: InterHub: A Naturalistic Trajectory Dataset with Dense Interaction for Autonomous Driving},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Y64P57Z3}},
  note         = {Machine review of arXiv:2411.18302}
}
read the original abstract

The driving interaction-a critical yet complex aspect of daily driving-lies at the core of autonomous driving research. However, real-world driving scenarios sparsely capture rich interaction events, limiting the availability of comprehensive trajectory datasets for this purpose. To address this challenge, we present InterHub, a dense interaction dataset derived by mining interaction events from extensive naturalistic driving records. We employ formal methods to describe and extract multi-agent interaction events, exposing the limitations of existing autonomous driving solutions. Additionally, we introduce a user-friendly toolkit enabling the expansion of InterHub with both public and private data. By unifying, categorizing, and analyzing diverse interaction events, InterHub facilitates cross-comparative studies and large-scale research, thereby advancing the evaluation and development of autonomous driving technologies.

Figures

Figures reproduced from arXiv: 2411.18302 by the authors.

Figure 1
Figure 1. Road map of dense driving interaction dataset construction and application. from most of the driving-interaction-related studies rely heavily on interaction clips which are usually from a single dataset. This dependence, coupled with the inherent limitations and lack of diversity in a single dataset, can lead to conclusions that may not be broadly applicable across different contexts. For instance, Li et al.33 noted… view at source ↗
Figure 2
Figure 2. Accumulated usage of widely adopted naturalistic driving datasets. Related works Recent years witnessed substantial exploration in the field of driving interaction where the public naturalistic driving datasets contribute a momentum [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Overall data structure of InterHub lane connectivity (i.e., lanes accessible from the left or right) and successor/predecessor lanes (i.e., lanes continuing from or leading to the current one). Extracting interaction events using the formal method The diverse types and sparse distribution of driving interaction events, along with the lack of research on defining and quantifying their spatio-temporal scope, have made… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Interaction defined by min-effort conflict resolution. To navigate safely and orderly through conflict points, certain vehicles shall act to adjust their motion states. We consider their actions in the mean of longitudinal acceleration. Thus, given the initial scenario…
Figure 5
Figure 5. Figure 5: Formal description of interaction. time difference when they pass through this intersection point is less than ConfTime (a threshold set to 3 seconds), as determined by the function TimeDiffLimit. The function, Time2inter(x (obji ),IntersectPoint(traj1 ,traj2 )) , calc…
Figure 6
Figure 6. Figure 6: Dynamic process of a driving interaction case. Interacting agents (with most efforts assigned to them in resolving conflicts) are highlighted in red, while other agents are marked in blue. 134622 157651 95% 94% 4% 4% 188587 96% 3% 1% 487093 96 % 3% 1% 1% 1% 1.00 1.25 1…
Figure 7
Figure 7. Figure 7: Results of interaction extraction: (A) quantity of scenarios, (B) density of number of vehicles participating in multi-agent interaction events, (C) distribution of effective interaction duration of the driving interaction segments, (D) distribution of interaction inte…
Figure 8
Figure 8. Figure 8: Distributions of PET for different segments of driving interaction. Interaction duration Figure 7C illustrates that across all datasets, the effective interactions are temporally sparse. The duration of interactions is predominantly concentrated in the range of 0.25 to…
Figure 9
Figure 9. Figure 9: Planning scores over varied interaction intensity. • 2_case_visualize.py showcases the process of retrieving and visualizing typical interaction cases. Codes and data usage is restricted to research purposes only. Any commercial exploitation of the data requires separa…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. A knowledge-augmented dataset of high-risk driving scenarios with LLM annotations for autonomous driving

    cs.LG 2026-07 conditional novelty 6.0 of 10

    K-Risk curates 31,398 high-risk driving events from 20 trajectory datasets with multi-layered semantic and LLM-generated annotations validated via closed-loop simulation.

  2. Surprise Potential as a Measure of Interactivity in Driving Scenarios

    cs.RO 2025-02 conditional novelty 5.0 of 10

    A counterfactual surprise metric, Hist-prim with query-centric feedforward prediction and Wasserstein distance, identifies interactive driving scenarios with 0.82+ Spearman correlation to a human-trained reward model.

Reference graph

Works this paper leans on

51 extracted references · 46 canonical work pages · cited by 2 Pith papers

  1. [1]

    & Tsotsos, J

    Rasouli, A. & Tsotsos, J. K. Autonomous vehicles that interact with pedestrians: A survey of theory and practice. IEEE Trans. Intell. Transp. Syst.21, 900–918 (2020)

    work page 2020

  2. [2]

    Pelikan, H. R. M. Why autonomous driving is so hard: The social dimension of traffic. In Companion of the 2021 ACM/IEEE International Conference on Human-Robot Interaction, 81–85 (ACM, 2021)

    work page 2021

  3. [3]

    Koh, L. Y . & Yuen, K. F. Public acceptance of autonomous vehicles: Examining the joint influence of perceived vehicle performance and intelligent in-vehicle interaction quality.Transportation Research Part A: Policy and Practice178, 103864 (2023)

    work page 2023

  4. [4]

    & Bajcsy, R

    Driggs-Campbell, K., Govindarajan, V . & Bajcsy, R. Integrating intuitive driver models in autonomous planning for interactive maneuvers. IEEE Trans. Intell. Transp. Syst.18, 3461–3472 (2017)

    work page 2017

  5. [5]

    Chen, X. et al. FollowNet: A comprehensive benchmark for car-following behavior modeling. Sci Data 10, 828 (2023)

    work page 2023

  6. [6]

    Zhou, H., Ma, K., Liang, S.,Li, X. & Qu, X. A unified longitudinal trajectory dataset for automated vehicle. Sci Data 11, 1123 (2024)

    work page 2024

  7. [7]

    & Wang, M

    Zhang, T., Fu, M., Song, W., Yang, Y . & Wang, M. Trajectory prediction based on constraints of vehicle kinematics and social interaction. In 2020 IEEE International Conference on Systems, Man, and Cybernetics (SMC), 3957–3963 (IEEE, 2020)

    work page 2020

  8. [8]

    & Huh, K

    Min, K., Kim, H., Park, J., Kim, D. & Huh, K. Interaction aware trajectory prediction of surrounding vehicles with interaction network and deep ensemble. In 2020 IEEE Intelligent Vehicles Symposium (IV), 1714–1719 (IEEE, 2020)

    work page 2020

Show all 51 references
  1. [9]

    & Shen, S

    Ding, W., Chen, J. & Shen, S. Predicting vehicle behaviors over an extended horizon using behavior interaction network. Preprint at https://arxiv.org/abs/1903.00848 (2019)

    2019 arXiv

  2. [10]

    Interaction-aware decision-making for autonomous vehicles

    Chen, Y .et al. Interaction-aware decision-making for autonomous vehicles. IEEE Trans. Transp. Electrific.9, 4704–4715 (2023)

    2023

  3. [11]

    Huang, Z., Liu, H., Wu, J., Huang, W. & Lv, C. Learning interaction-aware motion prediction model for decision-making in autonomous driving. In 2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC), 4820–4826 (IEEE, 2023)

    2023

  4. [12]

    Interactive prediction and decision-making for autonomous vehicles: Online active learning with traffic entropy minimization

    Zhang, Y .et al. Interactive prediction and decision-making for autonomous vehicles: Online active learning with traffic entropy minimization. IEEE Trans. Intell. Transp. Syst.25, 17718–17732 (2024)

    2024

  5. [13]

    V ., Li, S., Pirani, M

    Shu, K., Mehrizi, R. V ., Li, S., Pirani, M. & Khajepour, A. Human inspired autonomous intersection handling using game theory. IEEE Trans. Intell. Transp. Syst.24, 11360–11371 (2023). 13/15

    2023

  6. [14]

    & Bowden, R

    Arbabi, S., Tavernini, D., Fallah, S. & Bowden, R. Planning for autonomous driving via interaction-aware probabilistic action policies. IEEE Access 10, 81699–81712 (2022)

    2022

  7. [15]

    Li, J. et al. An interaction-aware predictive motion planner for unmanned ground vehicles in dynamic street scenarios. Int. J. Rob. Autom. 34, (2019)

    2019

  8. [16]

    E., Kolmanovsky, I

    Liu, K., Li, N., Tseng, H. E., Kolmanovsky, I. & Girard, A. Interaction-aware trajectory prediction and planning for autonomous vehicles in forced merge scenarios. IEEE Trans. Intell. Transp. Syst.24, 474–488 (2023)

    2023

  9. [17]

    Safety-balanced driving-style aware trajectory planning in intersection scenarios with uncertain environment

    Wang, X.et al. Safety-balanced driving-style aware trajectory planning in intersection scenarios with uncertain environment. IEEE Trans. Intell. Veh.8, 2888–2898 (2023)

    2023

  10. [18]

    & Modorcea, L

    Paz, A., Veeramisti, N., Khaddar, R., de la Fuente-Mella, H. & Modorcea, L. Traffic and driving simulator based on architecture of interactive motion. Sci. World J. 2015, 340576 (2015)

    2015

  11. [19]

    & Yang, H

    Hasan, M., Perez, D., Shen, Y . & Yang, H. Distributed microscopic traffic simulation with human-in-the-loop enabled by virtual reality technologies. Adv. Eng. Softw.154, 102985 (2021)

    2021

  12. [20]

    Yang, X. et al. Real-virtual consistent traffic flow interaction. Graph. Models 106, 101048 (2019)

    2019

  13. [21]

    & Jin, X

    Chao, Q., Deng, Z. & Jin, X. Vehicle-pedestrian interaction for mixed traffic simulation.Comput. Animat. Virtual Worlds 26, 405–412 (2015)

    2015

  14. [22]

    & Sun, J

    Liu, Y ., Zhao, X. & Sun, J. Towards interactive autonomous vehicle testing: Vehicle-under-test-centered traffic simulation. Preprint at https://arxiv.org/abs/2406.02860 (2024)

    2024 arXiv

  15. [23]

    Zhang, Y ., Hang, P., Huang, C. & Lv, C. Human-like interactive behavior generation for autonomous vehicles: A Bayesian game-theoretic approach with Turing test. Adv. Intell. Syst. 4, 2100211 (2022)

    2022

  16. [24]

    Wei, Z. et al. Interactive critical scenario generation for autonomous vehicles testing based on in-depth crash data using reinforcement learning. IEEE Trans. Intell. Veh.1–12 (2024)

    2024

  17. [25]

    Houston, J. et al. One thousand and one hours: Self-driving motion prediction dataset. In Conference on Robot Learning, 409–418, https://woven.toyota/en/prediction-dataset (PMLR, 2021)

    2021

  18. [26]

    Ettinger, S. et al. Large scale interactive motion forecasting for autonomous driving: The Waymo Open Motion Dataset. In 2021 IEEE/CVF International Conference on Computer Vision (ICCV), 9690–9699, https://waymo.com/open/data/motion/ (IEEE, 2021)

    2021

  19. [27]

    Zhan, W. et al. INTERACTION dataset: An international, adversarial and cooperative motion dataset in interactive driving scenarios with semantic maps. Preprint at https://arxiv.org/abs/1910.03088 (2019)

    2019 arXiv

  20. [28]

    & Pavone, M

    Ivanovic, B., Song, G., Gilitschenski, I. & Pavone, M. trajdata: A unified interface to multiple human trajectory datasets. In Advances in Neural Information Processing Systems 36 (NEURIPS 2023) (eds. Oh, A. et al.) (Neural Information Processing Systems, 2023)

    2023

  21. [29]

    Ozkan, M. F. & Ma, Y . Socially compatible control design of automated vehicle in mixed traffic.IEEE Control Syst. Lett. 6, 1730–1735 (2022)

    2022

  22. [30]

    A method for analyzing the traffic process in a safety perspective

    Svensson, Å. A method for analyzing the traffic process in a safety perspective. Bulletin 166, Lund Institute of Technology, Lund University (1998)

    1998

  23. [31]

    Markkula, G. et al. Defining interactions: A conceptual framework for understanding interactive behaviour in human and automated road traffic. Theor. Issues Ergon. Sci.21, 728–752 (2020)

    2020

  24. [32]

    Madigan, R. et al. Understanding interactions between automated road transport systems and other road users: A video analysis. Transp. Res. Pt. F Traffic Psychol. Behav.66, 196–213 (2019)

    2019

  25. [33]

    L., Calvert, S

    Li, G., Jiao, Y ., Knoop, V . L., Calvert, S. C. & van Lint, J. W. C. Large car-following data based on Lyft level-5 open dataset: Following autonomous vehicles vs. human-driven vehicles. In 2023 26th International Conference on Intelligent Transportation Systems (ITSC), 5818–...

    2023

  26. [34]

    & Sun, J

    Hu, X., Zheng, Z., Chen, D. & Sun, J. Autonomous vehicle’s impact on traffic: Empirical evidence from Waymo open dataset and implications from modelling. IEEE Trans. Intell. Transp. Syst.24, 6711–6724 (2023)

    2023

  27. [35]

    & Jian, S

    Wen, X., Cui, Z. & Jian, S. Characterizing car-following behaviors of human drivers when following automated vehicles using the real-world dataset. Accid. Anal. Prev.172, 106689 (2022)

    2022

  28. [36]

    & Sun, L

    Wang, W., Wang, L., Zhang, C., Liu, C. & Sun, L. Social interactions for autonomous driving: A review and perspective. Preprint at https://arxiv.org/abs/2208.07541 (2022). 14/15

    2022 arXiv

  29. [37]

    & Wang, M

    Zhao, X., Sun, J. & Wang, M. Measuring sociality in driving interaction. IEEE Trans. Intell. Transp. Syst.(2024)

    2024

  30. [38]

    Li, G., Jiao, Y ., Calvert, S. C. & Lint, J. W. C. van. A conflict resolution dataset derived from Argoverse-2: Analysis of the safety and efficiency impacts of autonomous vehicles at intersections. Preprint at https://arxiv.org/abs/2308.13839 (2023)

    2023 arXiv

  31. [39]

    Wilson, B. et al . Argoverse 2: Next generation datasets for self-driving perception and forecasting. Preprint at https://arxiv.org/abs/2301.00493 (2023)

    2023 arXiv

  32. [40]

    S., Sukhai, M

    Azizi Soldouz, S., Hasnine, M. S., Sukhai, M. & Nurul Habib, K. Communication and mobility issues of visually impaired pedestrians with connected autonomous vehicles. Transp. Plan. Techn.46, 998–1023 (2023)

    2023

  33. [41]

    Cui, M. et al. Passing-yielding intention estimation during lane change conflict: A semantic-based Bayesian inference method. IET Intell. Transp. Syst. 17, 2285–2299 (2023)

    2023

  34. [42]

    E., Lee, J

    Domeyer, J. E., Lee, J. D., Toyoda, H., Mehler, B. & Reimer, B. Driver-pedestrian perceptual models demonstrate coupling: Implications for vehicle automation. IEEE Trans. Hum.-Mach. Syst. 52, 557–566 (2022)

    2022

  35. [43]

    & Althoff, M

    Maierhofer, S., Moosbrugger, P. & Althoff, M. Formalization of intersection traffic rules in temporal logic.IEEE Intell. Veh. Symp. 2022, 1135–1144 (2022)

    2022

  36. [44]

    Caesar, H. et al. NuPlan: A closed-loop ML-based planning benchmark for autonomous vehicles. Preprint at https://arxiv.org/abs/2106.11810 (2022)

    2022 arXiv

  37. [45]

    Li, G., Li, Z., Knoop, V . L. & Van Lint, H. Unravelling uncertainty in trajectory prediction using a non-parametric approach. Transp. Res. Part C Emerg. Technol.163, 104659 (2024)

    2024

  38. [46]

    & Schiele, B

    Shi, S., Jiang, L., Dai, D. & Schiele, B. Motion transformer with global intention localization and local movement refinement. In Advances in Neural Information Processing Systems 35 (NEURIPS 2022) (eds. Koyejo, S., et al.) (Neural Information Processing Systems, 2022)

    2022

  39. [47]

    & Karamouzas, I

    Xu, P., Hayet, J.-B. & Karamouzas, I. Context-aware timewise V AEs for real-time vehicle trajectory prediction.IEEE Robot. Autom. Lett. 8, 5440–5447 (2023)

    2023

  40. [48]

    L., Liniger, A., Schwarting, W., Rus, D

    Vazquez, J. L., Liniger, A., Schwarting, W., Rus, D. & Van Gool, L. Deep interactive motion prediction and planning: Playing games with motion prediction models. In Learning for Dynamics and Control Conference, V ol. 168 (JMLR, 2022)

    2022

  41. [49]

    & Pavone, M

    Salzmann, T., Ivanovic, B., Chakravarty, P. & Pavone, M. Trajectron++: Dynamically-feasible trajectory forecasting with heterogeneous data. Preprint at https://arxiv.org/abs/2001.03093 (2021)

    2021 arXiv

  42. [50]

    Huang, Z., Liu, H. & Lv, C. GameFormer: Game-theoretic modeling and learning of transformer-based interactive prediction and planning for autonomous driving. Preprint at https://arxiv.org/abs/2303.05760 (2023)

    2023 arXiv

  43. [51]

    Poggenhans, F. et al. Lanelet2: A high-definition map framework for the future of automated driving. In 2018 21st Int. Conf. Intell. Transp. Syst. (ITSC), 1672–1679 (IEEE, 2018). Acknowledgements This research is jointly sponsored by National Natural Science Foundation of Chin...

    2018

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.