Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

Sky-Drive: A Distributed Multi-Agent Simulation Platform for Human-AI Collaborative and Socially-Aware Future Transportation

T0 review · 4 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read Sky-Drive claims that a distributed, multimodal, human-in-the-loop simulation platform can turn mixed-traffic human-AI interaction research into a closed loop spanning scenario generation, data collection, algorithm training, and hardware…

desk verdict A plausible, honestly-written distributed simulation platform paper whose central latency claim is unmeasured and whose validation tables are recycled from prior work—worth serious revision, not a desk reject. read the letter →

arxiv 2504.18010 v2 pith:3C7HWSLG submitted 2025-04-25 cs.RO cs.AIcs.HC

Zilin Huang , Zihao Sheng , Zhengyang Wan , Yansong Qu , Yuhao Luo , Boyue Wang , Pei Li , Yen-Jung Chen
show 5 more authors
Jiancong Chen Keke Long Jiayi Meng Yue Leng Sikai Chen
This is my paper · ORCID
classification cs.ROcs.AIcs.HC
keywords DrivingSimulatorAutonomousVehiclesHuman-AICollaborationMulti-AgentSimulationDigitalTwinSocially-AwareHuman-in-the-LoopDistributed
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

Sky-Drive is an open-source simulation platform whose central claim is that autonomous-driving research needs to study mixed traffic—autonomous vehicles interacting with human drivers, pedestrians, and cyclists—inside a shared virtual world, and that this requires more than a single-machine simulator. The paper builds a distributed multi-agent architecture on top of a widely used open urban driving simulator so that different agents can be controlled from separate terminals while the world state stays synchronized. It adds a multi-modal human-in-the-loop layer that records gaze, voice, facial expression, heart-rate signals, and steering inputs, plus a two-way human-AI mentoring mechanism in which humans shape AI policies through takeovers and feedback while AI coaches human drivers. If these pieces work as described, researchers gain a reusable environment for collecting safety-critical interaction data, training socially aware driving policies, replaying real accidents, and eventually validating algorithms on physical vehicles. The paper demonstrates the platform on several training methods and reports performance tables for each.

What carries the argument

The load-bearing object is the RPC-over-simulator distributed layer: a host-client setup in which one terminal runs the global environment and each client terminal injects control for its own agent, with dual-port TCP over LAN/VLAN keeping states in sync and a web-service log for all streams. A second mechanism is the bidirectional mentoring loop formed by HAIM and AIHM: HAIM converts human takeovers and multimodal feedback into preference comparisons that shape a reward-free reinforcement-learning objective, while AIHM generates reference trajectories and adjusts scenario difficulty from driver performance and physiological state. A third is the digital twin pipeline, which fuses multi-source real-world data into virtual scenes and replays accident trajectories extracted by an object tracker. These mechanisms together are what allow scenario generation, data collection, algorithm training, and hardware integration to close into one loop.

What would settle it

Run a controlled scaling experiment in Sky-Drive: start with a host plus one client on a single switch, record end-to-end position divergence between the host's ground-truth agent states and each client's rendered states, then add clients (2, 4, 8) and repeat over Wi-Fi, VLAN, and across-site links. If mean position error grows beyond a small fraction of a vehicle length (e.g., >0.5 m) or per-step latency exceeds the simulation tick budget at any realistic configuration, the synchronized-distributed-simulation claim fails as stated; publishing the 0.3-ms figure with its exact conditions would also settle whether the reported latency is meaningful.

Watch

Extended reading notes

Core claim

Sky-Drive's claim is that a driving simulator can be turned into a distributed multi-terminal research instrument rather than a single-process benchmark. The platform extends an established open urban driving simulator with a remote-procedure-call (RPC) distributed layer in which one host terminal maintains the global scene and client terminals independently control autonomous vehicles, human-driven vehicles, and pedestrians; a dual-port TCP system over LAN or VLAN keeps the terminals synchronized, and a real-time web monitor streams positions, velocities, video, and sensor data to a central log. On top of this shared world, Sky-Drive layers a multi-modal human-in-the-loop framework (a VR headset with eye tracking, voice transcription, an in-cabin camera, a smartwatch for physiological signals, and a racing wheel) and a human-AI collaboration mechanism with two modules: the human-as-AI-mentor module, which treats takeovers and preference feedback as implicit policy corrections in a reward-free reinforcement-learning objective, and the AI-as-human-mentor module, which uses physics-enhanced residual learning to generate reference trajectories and personalized training scenarios. A digital twin framework ingests high-definition maps, traffic-camera feeds, and historical records to reconstruct real environments, and an accident replay pipeline tracks objects from real crash videos and replays them for analysis. The reported experiments place Sky-Drive as the substrate for human-feedback RL, vision-language-model-reward RL, personalized curriculum learning, and accident replay.

Load-bearing premise

The central load-bearing premise is that the RPC-based host-client setup can keep the global simulation state synchronized across terminals at interactive latency; the paper reports a 0.3-ms low-latency figure without stating the measurement conditions, number of terminals, or network hardware, so if state consistency degrades as agents or sites are added, the distributed, human-in-the-loop, and collaborative-training claims all rest on unverified ground.

Editorial extensions

If this is right

  • Mixed-traffic interaction studies can move from single-machine rule-based traffic to live human participants controlling AVs, HVs, and pedestrians from separate terminals in one synchronized world.
  • Human takeovers become training signal: the HAIM formulation lets an AV policy be refined toward human preferences with roughly 60 times fewer samples than PPO (8,000 versus 500,000 in the reported comparison).
  • AIHM turns the same platform into a driver coach: real-time reference trajectories, annotated replays, and difficulty-adjusted scenario generation are driven by measured performance and physiological state.
  • Real-world accident footage can be replayed in simulation via tracked object trajectories, providing repeatable safety-critical scenarios for training and post-incident analysis.
  • The digital twin pipeline connects live traffic data feeds to the simulator, allowing scenario generation and evaluation to track a real road segment's current traffic state.

Reading between the lines

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

  • If the distributed synchronization holds at scale, Sky-Drive's architecture would enable multi-site human-subject studies where participants in different cities interact in the same traffic scene, which could expose regional differences in yielding norms and right-of-way conventions—something single-site simulators cannot measure.
  • The 0.3-millisecond latency claim is almost certainly a best-case single-switch figure; a natural extension is a public benchmark reporting state divergence and latency as functions of terminal count, agent count, and network path, without which the distributed claim lacks a quantitative envelope.
  • The preference-from-takeover mechanism generalizes beyond driving: the same HAIM/AIHM loop could turn Sky-Drive into a general-purpose human-AI alignment testbed for teleoperated robots, delivery drones, or assistive mobility devices, where corrective interventions are cheap to collect.
  • The accident replay pipeline, combined with a counterfactual AV control policy, suggests a concrete safety-validation workflow: replay real crashes, swap in a candidate AV policy, and measure whether the collision is avoided—an evaluation not possible from static footage.
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

4 major / 5 minor

Summary. The paper presents Sky-Drive, an open-source simulation platform built on CARLA that claims four core innovations: (a) a distributed multi-agent architecture using RPC for synchronized simulation across terminals, (b) a multi-modal human-in-the-loop framework with VR, eye tracking, voice, facial expression, physiological, and steering inputs, (c) a human-AI collaboration mechanism with HAIM (human as AI mentor) and AIHM (AI as human mentor) modules, and (d) a digital twin framework for reconstructing real-world transportation scenes. The paper also describes planned foundation-model and hardware-in-the-loop integrations. Section V presents five use cases: VR-based AV-HRU interaction, HAIM-based deep RL, VLM-enabled RL, personalized safety-critical curriculum learning, and accident data replay. The platform is positioned as a bridge between scenario generation, data collection, algorithm training, and hardware integration for socially-aware autonomous driving research.

Significance. If the distributed architecture and human-in-the-loop capabilities work as described, Sky-Drive would be a useful open-source infrastructure for studying mixed-traffic interactions with real human participants across terminals, an area where existing simulators are weak. The paper's strengths are its broad and clearly organized system description, a plausible modular design, and the authors' prior work on HAIM-DRL, VLM-RL, and CurricuVLM, which provide partial evidence that the individual algorithmic components have value. The paper also transparently marks several components (foundation models, HIL, multi-location VLAN testing) as planned rather than demonstrated. The central weakness is that the platform's own performance and synchronization claims are not measured, and the reported quantitative validations are recycled from earlier papers rather than produced inside Sky-Drive; this leaves the paper's core contribution unvalidated as a systems contribution.

major comments (4)
  1. [Section IV-A.2] The claim that Sky-Drive achieves "low latency of 0.3 milliseconds" for real-time interactions is asserted without any measurement protocol. The paper does not report the number of terminals, network hardware, payload size, end-to-end control latency, CARLA tick alignment, or synchronization error between client-side states. Because the entire distributed multi-agent premise depends on consistent global state across terminals, this figure must be replaced with actual measurements: latency distribution, state divergence (e.g., position/velocity error between host and clients), and scaling behavior as the number of agents and terminals grows. As written, the central performance claim is unsupported.
  2. [Section V, Tables II-IV] The quantitative validations for HAIM-DRL, VLM-RL, and CurricuVLM are explicitly recycled: the table notes state "The results are based on data reported in [9]" and similar for [57] and [63]. These tables therefore do not validate Sky-Drive itself; they only show that the authors' earlier algorithms perform well in their respective previous experimental settings. The manuscript needs at least one in-platform experiment that exercises Sky-Drive's distributed multi-terminal synchronization and human-in-the-loop data collection, reporting metrics such as synchronization error, interaction latency, and participant throughput. Without such an experiment, the claim that Sky-Drive "enables" these use cases remains an assertion.
  3. [Section IV-A.1] CARLA already supports multiple RPC clients connecting to a single server. The paper does not clearly explain what Sky-Drive's rpclib-based extension adds beyond CARLA's native multi-client mode. To substantiate the novelty of the "distributed multi-agent architecture," the authors should specify the synchronization mechanism (e.g., who owns the authoritative state, how client-side predictions are reconciled with server ticks, how conflicts are resolved) and describe a concrete scenario where CARLA's native mode fails and Sky-Drive succeeds. Without this, the distributed architecture claim is not differentiated from existing CARLA functionality.
  4. [Section V-A] The VR-based AV-HRU interaction case study describes an experimental setup but reports no results: there are no participant counts, no gaze or reaction-time statistics, and no yielding or near-miss outcomes. As presented, this is a demonstration of a planned procedure, not a validation. The authors should either report results from a pilot study or explicitly label this subsection as an illustrative setup rather than an evaluation.
minor comments (5)
  1. [Section IV-A.3] The subsection title contains a typo: "Real-time Monitoring Webstie" should be "Real-time Monitoring Website."
  2. [Section I] The phrase "A Vs" is inconsistently spaced in the introduction (e.g., "A Vs, HVs, and pedestrians" vs. "AVs" elsewhere); please standardize the notation.
  3. [Section IV-A.2] "Sky-Drive utilize" should be "Sky-Drive utilizes" for subject-verb agreement.
  4. [Table I] The table note defines "Distributed Multi-agent Simulation" but the caption could also clarify that all other columns are binary capabilities; consider adding a legend for the checkmark/dash symbols.
  5. [Section V-B] Equation (3) is stated to be "specifically designed as [9]" but the parameters α, β, φ are not defined in the text; please provide their roles or at least a pointer to the original paper for these hyperparameters.

Circularity Check

2 steps flagged · score 4.0 of 10

Validation of the HAIM, VLM-RL, and CurricuVLM use cases rests on the authors' own prior papers [9], [57], [63] rather than on in-platform measurements, making the demonstrated-effectiveness claims partially self-citational.

  1. self citation load bearing [Section V.B 'HAIM-based Deep Reinforcement Learning', Table II note (pp. 8-9)]
    "As evidenced by Tab. II, the HAIM-DRL was successfully implemented and evaluated within the Sky-Drive platform... These results validate Sky-Drive's capability to support closed-loop human-AI training... Note: The results are based on data reported in [9]. For detailed definitions of evaluation metrics and descriptions of baseline methods, please refer to the original paper."

    The only quantitative evidence offered for the HAIM use case is Table II, which the paper itself labels as data from [9]—the authors' own HAIM-DRL paper. The surrounding prose converts that recycled table into a demonstration that HAIM training was 'implemented and evaluated within the Sky-Drive platform,' but no Sky-Drive-specific measurement, synchronization metric, or multi-terminal run is reported for this experiment. The validation of Sky-Drive's HAIM capability thus reduces to a self-citation: the claim is supported by [9], not by any new in-platform result. This is load-bearing because this use case is one of the paper's four claimed demonstrations.

  2. self citation load bearing [Section V.C 'Vision Language Model-Enabled Reinforcement Learning' and Section V.D 'Personalized Safety-Critical Curriculum Learning', Tables III-IV notes]
    "The successful implementation of VLM-RL within the Sky-Drive platform validates its capability to support large-scale, multimodal policy learning... Note: The results are based on data reported in [57]... Note: The results are based on data reported in [63]."

    As with the HAIM case, the platform-capability claims for VLM-RL and CurricuVLM are evidenced by tables whose notes explicitly attribute the numbers to the authors' own prior works [57] and [63]. The prose asserts that the algorithms were 'successfully implemented within the Sky-Drive platform,' but the data is not generated by or measured on Sky-Drive in this paper. The demonstration of two more headline capabilities therefore rests on self-citations rather than on in-platform evaluation. Together with Section V.B, this makes the paper's 'effectiveness demonstrated through multiple application cases' conclusion largely self-referential.

full rationale

Sky-Drive is a systems paper, not a mathematical derivation, so most of its content is architectural description rather than a claim chain. I found no self-definitional equations, no fitted parameter renamed as a prediction, and no uniqueness theorem imported from the authors' prior work. The central circularity concern is narrower: the paper's proof-of-concept sections for HAIM, VLM-RL, and CurricuVLM reuse quantitative tables from the authors' own prior papers [9], [57], [63], while the surrounding text says these results validate Sky-Drive itself. That is a load-bearing self-citation pattern for the 'demonstrated effectiveness' claim, because the tables are explicitly not new in-platform measurements. However, the platform has independent substance: the distributed RPC architecture, the open-source code release, the digital twin pilot description, the accident replay framework, and the hardware integration plans are all separable from the recycled algorithm benchmarks. The unverified 0.3 ms latency figure and the absence of synchronization-error measurements are serious evidence gaps, but they are not circularity under the hard rules, so they do not raise the score. Overall, this is a case of partial circularity in the validation narrative rather than a derivation that reduces to its inputs by construction, hence a score of 4.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

Sky-Drive's central contribution is an assembly of existing components. The uncharged assumptions above are the engineering and behavioral premises that must hold for the platform's claimed utility to be real. No new physical entity, force, dimension, or parameter-free derivation is introduced.

free parameters (3)
  • Preference-RL objective weights (alpha, beta, phi) = Not specified; inherited from HAIM-DRL [9]
    Eq. (3) uses these weights for entropy regularization, takeover penalty, and disturbance penalty. No values or tuning procedure are reported in this paper.
  • VLM semantic reward weights (alpha, beta) = Not specified; inherited from VLM-RL [57]
    Eq. (4) combines positive and negative CLIP similarities with weights alpha and beta. No values are given here.
  • Reward synthesis factors (f_center, f_angle, f_stability, v_target) = Not specified; hand-designed in VLM-RL [57]
    Eq. (5) multiplies hand-crafted state-dependent terms. These are chosen functions, not fitted here, but they affect the demonstration results.
assumptions (6)
  • domain assumption CARLA is a sufficiently faithful base simulator for vehicle dynamics, sensors, and traffic interaction.
    The entire platform is built on CARLA (Section IV). No fidelity comparison to real traffic or to other simulators is provided.
  • domain assumption Multi-modal signals (gaze, voice, facial expression, heart rate) can be synchronized in real time and reliably indicate driver state or preference.
    Section IV-B describes the sensors, but no calibration, synchronization error, or validation of inferred states is reported.
  • domain assumption Human takeovers and interventions are valid implicit preference labels for training AV policies.
    HAIM-DRL in Section V-B treats takeovers as rejection signals and uses them in Eq. (2). The reliability of this assumption is not tested in this paper.
  • domain assumption IDM and MOBIL are valid physics-informed priors for socially compliant driving.
    Section IV-C.1 imports these traffic-flow models as domain knowledge. No sensitivity analysis is provided.
  • domain assumption Digital-twin data sources can be aligned and reconstructed into a faithful CARLA scene.
    Section IV-D describes temporal alignment and mapping, but no quantitative fidelity metric is reported.
  • domain assumption CenterTrack 2D-to-3D reconstruction preserves accident dynamics well enough for replay.
    Section V-E uses CenterTrack for accident replay. The paper mentions a quality-assessment module but reports no scores.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Sky-Drive: A Distributed Multi-Agent Simulation Platform for Human-AI Collaborative and Socially-Aware Future Transportation." pith.science (2026). https://pith.science/paper/3C7HWSLG

@misc{pith2026250418010,
  author       = {Pith},
  title        = {Pith review of: Sky-Drive: A Distributed Multi-Agent Simulation Platform for Human-AI Collaborative and Socially-Aware Future Transportation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3C7HWSLG}},
  note         = {Machine review of arXiv:2504.18010}
}
read the original abstract

Recent advances in autonomous system simulation platforms have significantly enhanced the safe and scalable testing of driving policies. However, existing simulators do not yet fully meet the needs of future transportation research-particularly in enabling effective human-AI collaboration and modeling socially-aware driving agents. This paper introduces Sky-Drive, a novel distributed multi-agent simulation platform that addresses these limitations through four key innovations: (a) a distributed architecture for synchronized simulation across multiple terminals; (b) a multi-modal human-in-the-loop framework integrating diverse sensors to collect rich behavioral data; (c) a human-AI collaboration mechanism supporting continuous and adaptive knowledge exchange; and (d) a digital twin framework for constructing high-fidelity virtual replicas of real-world transportation environments. Sky-Drive supports diverse applications such as autonomous vehicle-human road users interaction modeling, human-in-the-loop training, socially-aware reinforcement learning, personalized driving development, and customized scenario generation. Future extensions will incorporate foundation models for context-aware decision support and hardware-in-the-loop testing for real-world validation. By bridging scenario generation, data collection, algorithm training, and hardware integration, Sky-Drive has the potential to become a foundational platform for the next generation of human-centered and socially-aware autonomous transportation systems research. The demo video and code are available at:https://sky-lab-uw.github.io/Sky-Drive-website/

Figures

Figures reproduced from arXiv: 2504.18010 by the authors.

Figure 1
Figure 1. Overview of Sky-Drive’s key components and functionalities. (a) a distributed multi-agent architecture enabling synchronized simulation across multiple [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Workflow of Sky-Drive. (a) scenario generation & data collection through CARLA-based synthetic environments and digital twin integration of real [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Illustration of Sky-Drive’s distributed multi-agent architecture. Sky-Drive enables synchronized simulation across multiple terminals while maintaining [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: VR-based experimental setup for studying AV-HRU interactions at [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Qualitative examples. Each scenario is downsampled to four frames [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 7
Figure 7. Figure 7: LLM-based system enabling personalized autonomous driving. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]

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. Simulation for All: A Step-by-Step Cookbook for Developing Human-Centered Multi-Agent Transportation Simulators

    cs.MA 2025-07 conditional novelty 6.0 of 10

    A step-by-step cookbook and open-source codebase for building a multi-agent VR transportation simulator where pedestrians, cyclists, drivers, and transit users interact in real time while their physiological and neura...

  2. A Survey of World Models for Autonomous Driving

    cs.RO 2025-01 conditional novelty 2.0 of 10

    A survey presenting a three-branch taxonomy of world models for autonomous driving, plus benchmark tables comparing representative generation and planning methods on nuScenes, Waymo, Occ3D, and CarlaSC.

Reference graph

Works this paper leans on

75 extracted references · 62 canonical work pages · cited by 2 Pith papers

  1. [9]

    Human as ai mentor: En- hanced human-in-the-loop reinforcement learning for safe and efficient autonomous driving,

    Z. Huang, Z. Sheng, C. Ma, and S. Chen, “Human as ai mentor: En- hanced human-in-the-loop reinforcement learning for safe and efficient autonomous driving,” Commun. Transp. Res. , vol. 4, p. 100127, 2024

    work page 2024

  2. [57]

    Vlm-rl: A unified vision language models and reinforcement learning framework for safe autonomous driving,

    Z. Huang, Z. Sheng, Y . Qu, J. You, and S. Chen, “Vlm-rl: A unified vision language models and reinforcement learning framework for safe autonomous driving,” arXiv preprint arXiv:2412.15544 , 2024

    arXiv 2024

  3. [63]

    Curricuvlm: Towards safe autonomous driving via personalized safety- critical curriculum learning with vision-language models,

    Z. Sheng, Z. Huang, Y . Qu, Y . Leng, S. Bhavanam, and S. Chen, “Curricuvlm: Towards safe autonomous driving via personalized safety- critical curriculum learning with vision-language models,”arXiv preprint arXiv:2502.15119, 2025

    arXiv 2025

  4. [1]

    Convergence of emerging transportation trends: A comprehensive review of shared autonomous vehicles,

    D. Almaskati, S. Kermanshachi, and A. Pamidimukkala, “Convergence of emerging transportation trends: A comprehensive review of shared autonomous vehicles,” J. Intell. Connected Veh., vol. 7, no. 3, pp. 177– 189, 2024

    work page 2024

  5. [2]

    Kinematics-aware multigraph attention network with residual learning for heterogeneous trajectory prediction,

    Z. Sheng, Z. Huang, and S. Chen, “Kinematics-aware multigraph attention network with residual learning for heterogeneous trajectory prediction,” J. Intell. Connected Veh. , vol. 7, no. 2, pp. 138–150, 2024

    work page 2024

  6. [3]

    A review of vehicle detection methods based on computer vision,

    C. Ma and F. Xue, “A review of vehicle detection methods based on computer vision,” J. Intell. Connected Veh., vol. 7, no. 1, pp. 1–18, 2024

    work page 2024

  7. [4]

    A taxon- omy for autonomous vehicles considering ambient road infrastructure,

    S. Chen, S. Zong, T. Chen, Z. Huang, Y . Chen, and S. Labi, “A taxon- omy for autonomous vehicles considering ambient road infrastructure,” Sustainability, vol. 15, no. 14, p. 11258, 2023

    work page 2023

  8. [5]

    Modular flying vehicles: Scheduling modes, social benefits, and challenges,

    D. Lv, Y . Wang, L. Wang, Y . Fei, K. Wang, and X. Qu, “Modular flying vehicles: Scheduling modes, social benefits, and challenges,” p. 100144, 2024

    work page 2024

Show all 75 references
  1. [6]

    Towards socially responsive autonomous vehicles: A reinforcement learning framework with driving priors and coordination awareness,

    J. Liu, D. Zhou, P. Hang, Y . Ni, and J. Sun, “Towards socially responsive autonomous vehicles: A reinforcement learning framework with driving priors and coordination awareness,” IEEE Transactions on Intelligent Vehicles, vol. 9, no. 1, pp. 827–838, 2023

    2023

  2. [7]

    Driving to safety: How many miles of driving would it take to demonstrate autonomous vehicle reliability?

    N. Kalra and S. M. Paddock, “Driving to safety: How many miles of driving would it take to demonstrate autonomous vehicle reliability?” Transp. Res. Part A: Policy Pract. , vol. 94, pp. 182–193, 2016

    2016

  3. [8]

    Dense reinforcement learning for safety validation of autonomous vehicles,

    S. Feng, H. Sun, X. Yan, H. Zhu, Z. Zou, S. Shen, and H. X. Liu, “Dense reinforcement learning for safety validation of autonomous vehicles,” Nature, vol. 615, no. 7953, pp. 620–627, 2023

    2023

  4. [10]

    Carla: An open urban driving simulator,

    A. Dosovitskiy, G. Ros, F. Codevilla, A. Lopez, and V . Koltun, “Carla: An open urban driving simulator,” in Proc. Conf. Robot Learn. PMLR, 2017, pp. 1–16

    2017

  5. [11]

    Airsim: High-fidelity visual and physical simulation for autonomous vehicles,

    S. Shah, D. Dey, C. Lovett, and A. Kapoor, “Airsim: High-fidelity visual and physical simulation for autonomous vehicles,” in Proc. Int. Conf. Field Serv. Robot. Springer, 2018, pp. 621–635

    2018

  6. [12]

    Microscopic traffic simulation using sumo,

    P. A. Lopez, M. Behrisch, L. Bieker-Walz, J. Erdmann, Y .-P. Fl ¨otter¨od, R. Hilbrich, L. L ¨ucken, J. Rummel, P. Wagner, and E. Wießner, “Microscopic traffic simulation using sumo,” in 2018 21st international conference on intelligent transportation systems (ITSC) . Ieee, 20...

    2018

  7. [13]

    Ptv vissim,

    P. Group, “Ptv vissim,” https://www.ptvgroup.com/en/solutions/ ptv-vissim/, 2024, accessed: 2024-11-11

    2024

  8. [14]

    An environment for autonomous driving decision-making,

    E. Leurent, “An environment for autonomous driving decision-making,”

  9. [15]

    Metadrive: Composing diverse driving scenarios for generalizable reinforcement learning,

    Q. Li, Z. Peng, L. Feng, Q. Zhang, Z. Xue, and B. Zhou, “Metadrive: Composing diverse driving scenarios for generalizable reinforcement learning,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 45, no. 3, pp. 3461–3475, 2022

    2022

  10. [16]

    Smarts: An open-source scalable multi-agent rl training school for autonomous driving,

    M. Zhou, J. Luo, J. Villella, Y . Yang, D. Rusu, J. Miao, W. Zhang, M. Alban, I. Fadakar, Z. Chen et al., “Smarts: An open-source scalable multi-agent rl training school for autonomous driving,” in Proc. Conf. Robot Learn. PMLR, 2021, pp. 264–285

    2021

  11. [17]

    (2025) Carsim

    Mechanical Simulation. (2025) Carsim. [Online]. Available: https: //www.carsim.com

    2025

  12. [18]

    Automotive

    I. Automotive. (2025) Carmaker. [Online]. Available: https: //ipg-automotive.com/products-services/simulation-software/carmaker/

    2025

  13. [19]

    Waymax: An accelerated, data-driven simulator for large-scale autonomous driving research,

    C. Gulino, J. Fu, W. Luo, G. Tucker, E. Bronstein, Y . Lu, J. Harb, X. Pan, Y . Wang, X. Chenet al., “Waymax: An accelerated, data-driven simulator for large-scale autonomous driving research,” Adv. Neural Inf. Process. Syst., vol. 36, pp. 7730–7742, 2023

    2023

  14. [20]

    Sce- narionet: Open-source platform for large-scale traffic scenario simulation and modeling,

    Q. Li, Z. M. Peng, L. Feng, Z. Liu, C. Duan, W. Mo, and B. Zhou, “Sce- narionet: Open-source platform for large-scale traffic scenario simulation and modeling,” Adv. Neural Inf. Process. Syst. , vol. 36, pp. 3894–3920, 2023

    2023

  15. [21]

    Nocturne: a scalable driving benchmark for bringing multi-agent learning one step closer to the real world,

    E. Vinitsky, N. Lichtl ´e, X. Yang, B. Amos, and J. Foerster, “Nocturne: a scalable driving benchmark for bringing multi-agent learning one step closer to the real world,” Adv. Neural Inf. Process. Syst. , vol. 35, pp. 3962–3974, 2022

    2022

  16. [22]

    Gpt-4 technical report,

    OpenAI, “Gpt-4 technical report,” arXiv preprint arXiv:2303.08774 , 2023

    2023 arXiv

  17. [23]

    The llama 3 herd of models,

    A. Dubey, A. Jauhri, A. Pandey, A. Kadian, A. Al-Dahle, A. Letman, A. Mathur, A. Schelten, A. Yang, A. Fan et al. , “The llama 3 herd of models,” arXiv preprint arXiv:2407.21783 , 2024

    2024 arXiv

  18. [24]

    Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution,

    P. Wang, S. Bai, S. Tan, S. Wang, Z. Fan, J. Bai, K. Chen, X. Liu, J. Wang, W. Ge, Y . Fan, K. Dang, M. Du, X. Ren, R. Men, D. Liu, C. Zhou, J. Zhou, and J. Lin, “Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution,” arXiv preprint arXiv:2409....

    2024 arXiv

  19. [25]

    Gpt- 4 enhanced multimodal grounding for autonomous driving: Leveraging cross-modal attention with large language models,

    H. Liao, H. Shen, Z. Li, C. Wang, G. Li, Y . Bie, and C. Xu, “Gpt- 4 enhanced multimodal grounding for autonomous driving: Leveraging cross-modal attention with large language models,” Commun. Transp. Res., vol. 4, p. 100116, 2024

    2024

  20. [26]

    Drivearena: A closed-loop generative simulation platform for autonomous driving,

    X. Yang, L. Wen, Y . Ma, J. Mei, X. Li, T. Wei, W. Lei, D. Fu, P. Cai, M. Dou et al., “Drivearena: A closed-loop generative simulation platform for autonomous driving,” arXiv preprint arXiv:2408.00415 , 2024

    2024 arXiv

  21. [27]

    Chatscene: Knowledge-enabled safety- critical scenario generation for autonomous vehicles,

    J. Zhang, C. Xu, and B. Li, “Chatscene: Knowledge-enabled safety- critical scenario generation for autonomous vehicles,” in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. , 2024, pp. 15 459– 15 469

    2024

  22. [28]

    Editable scene simulation for autonomous driving via collaborative llm- agents,

    Y . Wei, Z. Wang, Y . Lu, C. Xu, C. Liu, H. Zhao, S. Chen, and Y . Wang, “Editable scene simulation for autonomous driving via collaborative llm- agents,” in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit., 2024, pp. 15 077–15 087

    2024

  23. [29]

    Development of deep-learning-based autonomous agents for low-speed maneuvering in unity,

    R. Berta, L. Lazzaroni, A. Capello, M. Cossu, L. Forneris, A. Pighetti, and F. Bellotti, “Development of deep-learning-based autonomous agents for low-speed maneuvering in unity,” J. Intell. Connected Veh. , vol. 7, no. 3, pp. 229–244, 2024

    2024

  24. [30]

    Social interactions for autonomous driving: A review and perspectives,

    W. Wang, L. Wang, C. Zhang, C. Liu, L. Sun et al., “Social interactions for autonomous driving: A review and perspectives,” Found. Trends Robot., vol. 10, no. 3-4, pp. 198–376, 2022

    2022

  25. [31]

    Congested traffic states in empirical observations and microscopic simulations,

    M. Treiber, A. Hennecke, and D. Helbing, “Congested traffic states in empirical observations and microscopic simulations,” Phys. Rev. E , vol. 62, no. 2, p. 1805, 2000

    2000

  26. [32]

    General lane-changing model mobil for car-following models,

    A. Kesting, M. Treiber, and D. Helbing, “General lane-changing model mobil for car-following models,” Transp. Res. Rec. , vol. 1999, no. 1, pp. 86–94, 2007

    1999

  27. [33]

    Choose your simulator wisely: A review on open-source simulators for autonomous driving,

    Y . Li, W. Yuan, S. Zhang, W. Yan, Q. Shen, C. Wang, and M. Yang, “Choose your simulator wisely: A review on open-source simulators for autonomous driving,” IEEE Trans. Intell. Veh., 2024

    2024

  28. [34]

    Lgsvl simulator: A high fidelity simulator for autonomous driving,

    G. Rong, B. H. Shin, H. Tabatabaee, Q. Lu, S. Lemke, M. Mo ˇzeiko, E. Boise, G. Uhm, M. Gerow, S. Mehta et al., “Lgsvl simulator: A high fidelity simulator for autonomous driving,” in Proc. IEEE Int. Conf. Intell. Transp. Syst. (ITSC) . IEEE, 2020, pp. 1–6

    2020

  29. [35]

    Nvidia Drive End-to-End Platform for Software-Defined Vehicles,

    Nvidia, “Nvidia Drive End-to-End Platform for Software-Defined Vehicles,” Online, 2024, accessed: Mar. 12, 2024. [Online]. Available: https://www.nvidia.com/en-us/self-driving-cars/

    2024

  30. [36]

    The World’s Most Accurate Simulation Environment,

    rFpro, “The World’s Most Accurate Simulation Environment,” Online, 2023, accessed: Mar. 12, 2024. [Online]. Available: https://rfpro.com

    2023

  31. [37]

    Deepdrive: a simulator that allows anyone with a pc to push the state-of-the-art in self-driving,

    D. Team, “Deepdrive: a simulator that allows anyone with a pc to push the state-of-the-art in self-driving,” 2019. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 14

    2019

  32. [38]

    Garchingsim: An autonomous driving simulator with photorealistic scenes and minimalist workflow,

    L. Zhou, Y . Song, Y . Gao, Z. Yu, M. Sodamin, H. Liu, L. Ma, L. Liu, H. Liu, Y . Liu et al., “Garchingsim: An autonomous driving simulator with photorealistic scenes and minimalist workflow,” in Proc. IEEE Int. Conf. Intell. Transp. Syst. (ITSC) . IEEE, 2023, pp. 4227–4232

    2023

  33. [39]

    Flow: A modular learning framework for mixed autonomy traffic,

    C. Wu, A. R. Kreidieh, K. Parvate, E. Vinitsky, and A. M. Bayen, “Flow: A modular learning framework for mixed autonomy traffic,”IEEE Trans. Robot., vol. 38, no. 2, pp. 1270–1286, 2021

    2021

  34. [40]

    Cityflow: A multi-agent reinforcement learning environment for large scale city traffic scenario,

    H. Zhang, S. Feng, C. Liu, Y . Ding, Y . Zhu, Z. Zhou, W. Zhang, Y . Yu, H. Jin, and Z. Li, “Cityflow: A multi-agent reinforcement learning environment for large scale city traffic scenario,” in Proc. World Wide Web Conf., 2019, pp. 3620–3624

    2019

  35. [41]

    Sim4cv: A photo-realistic simulator for computer vision applications,

    M. M ¨uller, V . Casser, J. Lahoud, N. Smith, and B. Ghanem, “Sim4cv: A photo-realistic simulator for computer vision applications,” Int. J. Comput. Vis., vol. 126, pp. 902–919, 2018

    2018

  36. [42]

    TORCS, the open racing car simulator,

    B. Wymann, E. Espi ´e, C. Guionneau, C. Dimitrakakis, R. Coulom, and A. Sumner, “TORCS, the open racing car simulator,” 2020, accessed: Mar. 12, 2024. [Online]. Available: https://sourceforge.net/projects/torcs/

    2020

  37. [43]

    Summit: A simulator for urban driving in massive mixed traffic,

    P. Cai, Y . Lee, Y . Luo, and D. Hsu, “Summit: A simulator for urban driving in massive mixed traffic,” inProc. IEEE Int. Conf. Robot. Autom. IEEE, 2020, pp. 4023–4029

    2020

  38. [44]

    Multi-agent connected autonomous driving using deep reinforcement learning,

    P. Palanisamy, “Multi-agent connected autonomous driving using deep reinforcement learning,” in Proc. Int. Joint Conf. Neural Netw. IEEE, 2020, pp. 1–7

    2020

  39. [45]

    Robotics

    O. Robotics. (2025) Gazebo. [Online]. Available: https://gazebosim.org/

    2025

  40. [46]

    (2025) Vehicle dynamics blockset

    Mathworks. (2025) Vehicle dynamics blockset. [Online]. Available: https://www.mathworks.com/products/vehicle-dynamics.html

    2025

  41. [47]

    NVIDIA DRIVE Sim,

    NVIDIA Corporation, “NVIDIA DRIVE Sim,” 2025, https://developer. nvidia.com/drive/simulation

    2025

  42. [48]

    Applied Intuition,

    Applied Intuition, Inc., “Applied Intuition,” 2025, https://www. appliedintuition.com/

    2025

  43. [49]

    MORAI Inc

    MORAI Inc., “MORAI Inc.” 2025, https://www.morai.ai/

    2025

  44. [50]

    Vista 2.0: An open, data-driven simulator for multimodal sensing and policy learning for autonomous vehicles,

    A. Amini, T.-H. Wang, I. Gilitschenski, W. Schwarting, Z. Liu, S. Han, S. Karaman, and D. Rus, “Vista 2.0: An open, data-driven simulator for multimodal sensing and policy learning for autonomous vehicles,” in Proc. IEEE Int. Conf. Robot. Autom. IEEE, 2022, pp. 2419–2426

    2022

  45. [51]

    Gamma: A general agent motion model for autonomous driving,

    Y . Luo, P. Cai, Y . Lee, and D. Hsu, “Gamma: A general agent motion model for autonomous driving,” IEEE Robot. Autom. Lett., vol. 7, no. 2, pp. 3499–3506, 2022

    2022

  46. [52]

    LINGO-1: Exploring Natural Language for Autonomous Driving,

    Wayve, “LINGO-1: Exploring Natural Language for Autonomous Driving,” 2023, https://wayve.ai/thinking/ lingo-natural-language-autonomous-driving/

    2023

  47. [53]

    Bidirectional planning for au- tonomous driving framework with large language model,

    Z. Ma, Q. Sun, and T. Matsumaru, “Bidirectional planning for au- tonomous driving framework with large language model,” Sensors, vol. 24, no. 20, p. 6723, 2024

    2024

  48. [54]

    Surrealdriver: Designing generative driver agent simulation framework in urban contexts based on large language model,

    Y . Jin, X. Shen, H. Peng, X. Liu, J. Qin, J. Li, J. Xie, P. Gao, G. Zhou, and J. Gong, “Surrealdriver: Designing generative driver agent simulation framework in urban contexts based on large language model,” arXiv preprint arXiv:2309.13193 , vol. 5, no. 7, p. 8, 2023

    2023 arXiv

  49. [55]

    Human-machine collaborative design for accelerated design of compact deep neural networks for autonomous driving,

    M. J. Shafiee, M. Nentwig, Y . Kassahun, F. Li, S. Bochkarev, A. Kamal, D. Dolson, S. Altintas, A. Virani, and A. Wong, “Human-machine collaborative design for accelerated design of compact deep neural networks for autonomous driving,” arXiv preprint arXiv:1909.05587 , 2019

    1909 arXiv

  50. [56]

    Robust speech recognition via large-scale weak super- vision,

    A. Radford, J. W. Kim, T. Xu, G. Brockman, C. McLeavey, and I. Sutskever, “Robust speech recognition via large-scale weak super- vision,” in Proc. Int. Conf. Mach. Learn. PMLR, 2023, pp. 28 492– 28 518

    2023

  51. [58]

    Trustworthy human-ai collaboration: Reinforcement learning with human feedback and physics knowledge for safe autonomous driving,

    Z. Huang, Z. Sheng, and S. Chen, “Trustworthy human-ai collaboration: Reinforcement learning with human feedback and physics knowledge for safe autonomous driving,” arXiv preprint arXiv:2409.00858 , 2024

    2024 arXiv

  52. [59]

    A physics enhanced residual learning (perl) framework for vehicle trajectory pre- diction,

    K. Long, Z. Sheng, H. Shi, X. Li, S. Chen, and S. Ahn, “A physics enhanced residual learning (perl) framework for vehicle trajectory pre- diction,” Commun. Transp. Res. , 2025

    2025

  53. [60]

    Traffic expertise meets residual rl: Knowledge-informed model-based residual reinforcement learning for cav trajectory control,

    Z. Sheng, Z. Huang, and S. Chen, “Traffic expertise meets residual rl: Knowledge-informed model-based residual reinforcement learning for cav trajectory control,” Commun. Transp. Res., vol. 4, p. 100142, 2024

    2024

  54. [61]

    Ego-planning-guided multi-graph convolutional network for het- erogeneous agent trajectory prediction,

    ——, “Ego-planning-guided multi-graph convolutional network for het- erogeneous agent trajectory prediction,” Comput.-Aided Civ. Infrastruct. Eng., vol. 39, no. 22, pp. 3357–3374, 2024

    2024

  55. [62]

    Talk2traffic: Interac- tive and editable traffic scenario generation for autonomous driving with multimodal large language model,

    Z. Sheng, Z. Huang, Y . Qu, Y . Leng, and S. Chen, “Talk2traffic: Interac- tive and editable traffic scenario generation for autonomous driving with multimodal large language model,” in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR) Workshops , 2025

    2025

  56. [64]

    Tracking objects as points,

    X. Zhou, V . Koltun, and P. Kr ¨ahenb¨uhl, “Tracking objects as points,” in Proc. Eur. Conf. Comput. Vis. Springer, 2020, pp. 474–490

    2020

  57. [65]

    Personalizing driver agent using large language models for driving safety and smarter human– machine interactions,

    Z. Xu, T. Chen, Z. Huang, Y . Xing, and S. Chen, “Personalizing driver agent using large language models for driving safety and smarter human– machine interactions,” IEEE Intell. Transp. Syst. Mag. , 2025

    2025

  58. [66]

    End-to-end learning of driving models from large-scale video datasets,

    H. Xu, Y . Gao, F. Yu, and T. Darrell, “End-to-end learning of driving models from large-scale video datasets,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR) , 2017, pp. 2174–2182

    2017

  59. [67]

    Dorothie: Spoken dialogue for handling unexpected situations in interactive autonomous driving agents,

    Z. Ma, B. VanDerPloeg, C.-P. Bara, H. Yidong, E.-I. Kim, F. Gervits, M. Marge, and J. Chai, “Dorothie: Spoken dialogue for handling unexpected situations in interactive autonomous driving agents,” arXiv preprint arXiv:2210.12511, 2022

    arXiv

  60. [68]

    Lmdrive: Closed-loop end-to-end driving with large language models,

    H. Shao, Y . Hu, L. Wang, G. Song, S. L. Waslander, Y . Liu, and H. Li, “Lmdrive: Closed-loop end-to-end driving with large language models,” in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. , 2024, pp. 15 120–15 130

    2024

  61. [69]

    Uncertainty-based traffic accident anticipation with spatio-temporal relational learning,

    W. Bao, Q. Yu, and Y . Kong, “Uncertainty-based traffic accident anticipation with spatio-temporal relational learning,” in Proc. ACM Multimedia Conf., May 2020

    2020

  62. [70]

    Dota: unsupervised detection of traffic anomaly in driving videos,

    Y . Yao, X. Wang, M. Xu, Z. Pu, Y . Wang, E. Atkins, and D. Crandall, “Dota: unsupervised detection of traffic anomaly in driving videos,” IEEE Trans. Pattern Anal. Mach. Intell. , 2022

    2022

  63. [71]

    Drivecot: Integrating chain-of-thought reasoning with end-to-end driving,

    T. Wang, E. Xie, R. Chu, Z. Li, and P. Luo, “Drivecot: Integrating chain-of-thought reasoning with end-to-end driving,” arXiv preprint arXiv:2403.16996, 2024

    2024 arXiv

  64. [72]

    Study on enhancement of 3GPP Support for 5G V2X Services,

    “Study on enhancement of 3GPP Support for 5G V2X Services,” 3rd Generation Partnership Project (3GPP), Technical Report TR 22.886, 2020, available at: https://www.3gpp.org/ftp/Specs/archive/22 series/22. 886/

    2020

  65. [73]

    Toward c-v2x enabled connected transportation system: Rsu-based cooperative localization framework for autonomous vehicles,

    Z. Huang, S. Chen, Y . Pian, Z. Sheng, S. Ahn, and D. A. Noyce, “Toward c-v2x enabled connected transportation system: Rsu-based cooperative localization framework for autonomous vehicles,” IEEE Trans. Intell. Transp. Syst., 2024

    2024

  66. [74]

    V2x-vlm: End-to-end v2x cooperative au- tonomous driving through large vision-language models,

    J. You, H. Shi, Z. Jiang, Z. Huang, R. Gan, K. Wu, X. Cheng, X. Li, and B. Ran, “V2x-vlm: End-to-end v2x cooperative au- tonomous driving through large vision-language models,” arXiv preprint arXiv:2408.09251, 2024

    2024 arXiv

  67. [2018]

    Available: https://github.com/eleurent/highway-env

    [Online]. Available: https://github.com/eleurent/highway-env

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.