AI交通科学家：大模型驱动的自主交通科研

宫晓燕; 戴星原; 李芮霖; 吕宜生

doi:10.11959/j.issn.2096-6652.202605

Chinese

您当前的位置：

首页 >

文章列表页 >

AI交通科学家：大模型驱动的自主交通科研

学术论文 | 更新时间：2026-04-27

- AI交通科学家：大模型驱动的自主交通科研
- AI transportation scientist: LLM-driven autonomous transportation research
- 智能科学与技术学报 2026年8卷第1期页码：33-46
- 作者机构：
  
  1.中国科学院自动化研究所多模态人工智能系统全国重点实验室，北京 100190
  2.道路交通安全管控技术国家工程研究中心，北京 100006
  3.山东交通学院，山东济南 250357
  4.澳门科技大学创新工程学院，澳门 999078
- 作者简介：
  
  [ "宫晓燕（1976- ），女，博士，中国科学院自动化研究所多模态人工智能系统全国重点实验室高级工程师，主要研究方向为复杂系统管理与控制、交通流预测、人工智能。" ]
  [ "戴星原（1993- ），男，博士，中国科学院自动化研究所多模态人工智能系统全国重点实验室副研究员，主要研究方向为平行交通、平行驾驶、强化学习。" ]
  [ "李芮霖（2002- ），女，山东交通学院硕士生，主要研究方向为交通信号管控及交通管理。" ]
  [ "吕宜生（1983- ），男，博士，中国科学院自动化研究所多模态人工智能系统全国重点实验室研究员，澳门科技大学创新工程学院教授，主要研究方向为智能交通、无人驾驶、人工智能。" ]
- 基金信息：
  
  国家自然科学基金项目(62271485;62303462);道路交通安全管控技术国家工程研究中心开放课题(2024GCZXKFKT11A);山东省交通运输厅科技计划项目(2024B70);山东高速集团有限公司科技计划项目(HS2023B044)
- DOI：10.11959/j.issn.2096-6652.202605
  中图分类号： TP39
- 收稿：2025-12-16，
  
  修回：2026-03-13，
  
  录用：2026-03-17，
  
  纸质出版：2026-03-15
- 稿件说明：
移动端阅览
宫晓燕,戴星原,李芮霖等.AI交通科学家：大模型驱动的自主交通科研[J].智能科学与技术学报,2026,08(01):33-46.

Gong Xiaoyan,Dai Xingyuan,Li Ruilin,et al.AI transportation scientist: LLM-driven autonomous transportation research[J].Chinese Journal of Intelligent Science and Technology,2026,08(01):33-46.
宫晓燕,戴星原,李芮霖等.AI交通科学家：大模型驱动的自主交通科研[J].智能科学与技术学报,2026,08(01):33-46. DOI： 10.11959/j.issn.2096-6652.202605.

Gong Xiaoyan,Dai Xingyuan,Li Ruilin,et al.AI transportation scientist: LLM-driven autonomous transportation research[J].Chinese Journal of Intelligent Science and Technology,2026,08(01):33-46. DOI： 10.11959/j.issn.2096-6652.202605.

摘要

城市交通系统正加速演化为社会物理信息系统（cyber-physical-social system，CPSS），无人车、无人机及各类智能体持续融入，导致系统复杂性、动态性与耦合性急剧增强。传统以人工为主的“作坊式”科研模式，难以实现对系统快速演化行为的深入解析与实时响应。为应对上述挑战，立足平行智能理论，提出融合大模型认知能力与多智能体协同优势的自主交通科研框架——AI交通科学家。该框架构建交互层、认知层、实验层、支撑层4层架构，并以动态路由为调度引擎，针对机理发现、策略验证与系统优化3类交通科研问题，通过动态路由自适应编排各层智能体协作链路，形成匹配不同任务需求的科研流程。框架贯通文献分析、问题识别、假设生成、仿真验证、策略生成、反馈优化的全链条人机协同闭环，不仅能实现复杂交通问题的自动化识别与高效求解，还能驱动交通系统规律的自主挖掘，以及管控策略的创新迭代、多场景验证与持续优化，为CPSS背景下交通科研的智能化转型提供可落地的技术范式。

Abstract

Urban transportation systems are rapidly evolving into CPSS (cyber-physical-social system)

driven by the continuous integration of autonomous vehicles

unmanned aerial vehicles

and diverse intelligent agents. This evolution has dramatically increased system complexity

dynamics

and coupling

rendering traditional human-centric research paradigms insufficient for timely understanding and response to fast-evolving system behaviors. To address these challenges

an autonomous framework called "AI Transportation Scientist" was proposed to revolutionize transportation research through parallel intelligence. The architecture leveraged a synergy between large language model and multi-agent system across four functional layers (interaction

cognitive

experimental

and support). At its core

a dynamic routing engine adaptively scheduled intelligent agents to tackle mechanism discovery

strategy validation

and system optimization. By implementing a full-chain collaborative closed loop—encompassing problem identification

simulation

and feedback optimization—the framework enabled the autonomous discovery of transportation laws and the continuous evolution of control strategies. This research establishes a scalable technical paradigm for advancing transportation science within CPSS environments

ensuring both efficient problem-solving and innovative strategy iteration.

关键词

Keywords

references

Wang F Y. The emergence of intelligent enterprises: from CPS to CPSS[J]. IEEE Intelligent Systems, 2010, 25(4): 85-88.

杨静, 王晓, 王雨桐, 等. 平行智能与CPSS: 三十年发展的回顾与展望[J]. 自动化学报, 2023, 49(3): 614-634.

Yang J, Wang X, Wang Y T, et al. Parallel intelligence and CPSS in 30 years: an ACP approach[J]. Acta Automatica Sinica, 2023, 49(3): 614-634.

Feng Z H, Xue R Q, Yuan L, et al. Multi-agent embodied AI: advances and future directions[PP]. V2. (2025-06-21)[2025-12-16]. arXiv: arXiv.2505.05108.

He Y X, Wang D W, Huang F H, et al. Aerial-ground integrated vehicular networks: a UAV-vehicle collaboration perspective[J]. IEEE Transactions on Intelligent Transportation Systems, 2024, 25(6): 5154-5169.

王飞跃, 缪青海. 人工智能驱动的科学研究新范式: 从AI4S到智能科学[J]. 中国科学院院刊, 2023, 38(4): 536-540.

Wang F Y, Miao Q H. Novel paradigm for AI-driven scientific research: from AI4S to intelligent science[J]. Bulletin of Chinese Academy of Sciences, 2023, 38(4): 536-540.

Wang F Y. Parallel control and management for intelligent transportation systems: concepts, architectures, and applications[J]. IEEE Transactions on Intelligent Transportation Systems, 2010, 11(3): 630-638.

Lv Y S, Duan Y J, Kang W W, et al. Traffic flow prediction with big data: a deep learning approach[J]. IEEE Transactions on Intelligent Transportation Systems, 2015, 16(2): 865-873.

Li L, Lv Y S, Wang F Y. Traffic signal timing via deep reinforcement learning[J]. IEEE/CAA Journal of Automatica Sinica, 2016, 3(3): 247-254.

Radford A, Narasimhan K, Salimans T, et al. Improving language understanding by generative pre-training[EB]. (2018-06-11) [2025-12-16].

Wooldridge M J. An introduction to multiagent systems[M]. 2nd ed. Chichester: John Wiley & Sons, 2009.

Brown T B, Mann B, Ryder N, et al. Language models are few-shot learners[C]//Proceedings of the 34th International Conference on Neural Information Processing Systems. New York: ACM, 2020: 1877-1901.

King R D, Whelan K E, Jones F M, et al. Functional genomic hypothesis generation and experimentation by a robot scientist[J]. Nature, 2004, 427(6971): 247-252.

Abolhasani M, Kumacheva E. The rise of self-driving labs in chemical and materials sciences[J]. Nature Synthesis, 2023, 2(6): 483-492.

Touvron H, Lavril T, Izacard G, et al. LLaMA: open and efficient foundation language models[PP]. V1. (2023-02-27)[2025-12-16]. arXiv: arXiv.2302.13971.

Beltagy I, Lo K, Cohan A. SciBERT: a pretrained language model for scientific text[C]//Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). Stroudsburg: ACL, 2019: 3613-3618.

Taylor R, Kardas M, Cucurull G, et al. Galactica: a large language model for science[PP]. V1. (2022-11-16)[2025-12-16]. arXiv: arXiv.2211.09085.

She F, Wang N, Wu H, et al. SciGPT: a large language model for scientific literature understanding and knowledge discovery[PP]. V1. (2025-09-09) [2025-12-16]. arXiv: arXiv.2509.08032.

Thoppilan R, De Freitas D, Hall J, et al. LaMDA: language models for dialog applications[PP]. V3. (2022-02-10)[2025-12-16]. arXiv: arXiv. 2201.08239.

OpenAI. GPT-4 technical report[R]. 2023.

Bosma M, Chi E, Ichter B, et al. Chain-of-thought prompting elicits reasoning in large language models[C]//Proceedings of the Advances in Neural Information Processing Systems. Massachusetts: MIT Press, 2022: 24824-24837.

Zhang K Q, Yang Z R, Başar T. Multi-agent reinforcement learning: a selective overview of theories and algorithms[PP]. V2. (2019-11-24)[2025-12-16]. arXiv: arXiv.1911.10635.

Park J S, O'Brien J, Cai C J, et al. Generative agents: interactive simulacra of human behavior[C]//Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology. New York: ACM, 2023: 1-22.

Richards B. Auto-GPT: An autonomous GPT-4 experiment[EB]. (2023-03-30) [2025-12-16].

Chen W Z, Su Y S, Zuo J W, et al. AgentVerse: facilitating multi-agent collaboration and exploring emergent behaviors[PP]. V3. (2023-10-23)[2025-12-16]. arXiv: arXiv.2308.10848.

Ge Y Q, Hua W Y, Ji J C, et al. OpenAGI: when LLM meets domain experts[C]//Proceedings of the Advances in Neural Information Processing Systems. Massachusetts: MIT Press, 2023: 5539-5568.

Li X, Wu R, Liu X, et al. SciAgent: A Unified Multi-Agent System for Generalistic Scientific Reasoning[PP]. V2. (2025-11-07) [2025-12-16], arXiv: arXiv.2511.08151.

Leong S X, Griesbach C E, Zhang R, et al. Steering towards safe self-driving laboratories[J]. Nature Reviews Chemistry, 2025, 9(10): 707-722.

King R D, Rowland J, Oliver S G, et al. The automation of science[J]. Science, 2009, 324(5923): 85-89.

Symonds J, Thakkar A, Borsos Z, et al. A-Lab: an autonomous laboratory for the accelerated discovery of new materials[J]. Nature, 2023, 624: 86-91.

Wang H C, Fu T F, Du Y Q, et al. Scientific discovery in the age of artificial intelligence[J]. Nature, 2023, 620(7972): 47-60.

王飞跃. 平行系统方法与复杂系统的管理和控制[J]. 控制与决策, 2004, 19(5): 485-489, 514.

Wang F Y. Parallel system methods for management and control of complex systems[J]. Control and Decision, 2004, 19(5): 485-489, 514.

王飞跃. 关于复杂系统的建模、分析、控制和管理[J]. 复杂系统与复杂性科学, 2006, 3(2): 26-34.

Wang F Y. On the modeling, analysis, control and management of complex systems[J]. Complex Systems and Complexity Science, 2006, 3(2): 26-34.

王飞跃. 计算实验方法与复杂系统行为分析和决策评估[J]. 系统仿真学报, 2004, 16(5): 893-897.

Wang F Y. Computational experiments for behavior analysis and decision evaluation of complex systems[J]. Journal of System Simulation, 2004, 16(5): 893-897.

吕宜生, 陈圆圆, 金峻臣, 等. 平行交通: 虚实互动的智能交通管理与控制[J]. 智能科学与技术学报, 2019, 1(1): 21-33.

Lyu Y S, Chen Y Y, Jin J C, et al. Parallel transportation: virtual-real interaction for intelligent traffic management and control[J]. Chinese Journal of Intelligent Science and Technology, 2019, 1(1): 21-33.

王晓, 张翔宇, 周锐, 等. 基于平行测试的认知自动驾驶智能架构研究[J]. 自动化学报, 2024, 50(2): 356-371.

Wang X, Zhang X Y, Zhou R, et al. An intelligent architecture for cognitive autonomous driving based on parallel testing[J]. Acta Automatica Sinica, 2024, 50(2): 356-371.

Batty M. Digital twins[J]. Environment and Planning B: Urban Analytics and City Science, 2018, 45(5): 817-820.

Batty M. Digital twins in city planning[J]. Nature Computational Science, 2024, 4(3): 192-199.

Liu H Q, Su W K, Li T, et al. Digital twin enhanced multi-agent reinforcement learning for large-scale mobile network coverage optimization[J]. ACM Transactions on Knowledge Discovery from Data, 2025, 19(1): 1-23.

Dasgupta S, Rahman M, Jon S. Harnessing digital twin technology for adaptive traffic signal control: improving signalized intersection performance and user satisfaction[J]. IEEE Internet of Things Journal, 2024, 11(22): 36596-36618.

Liao X W, Leng S P, Sun Y, et al. A digital-twin-based traffic guidance scheme for autonomous driving[J]. IEEE Internet of Things Journal, 2024, 11(22): 36829-36840.

Nie L S, Wang X J, Zhao Q L, et al. Digital twin for transportation big data: a reinforcement learning-based network traffic prediction approach[J]. IEEE Transactions on Intelligent Transportation Systems, 2024, 25(1): 896-906.

Thonhofer E, Sigl S, Fischer M, et al. Infrastructure-based digital twins for cooperative, connected, automated driving and smart road services[J]. IEEE Open Journal of Intelligent Transportation Systems, 2023, 4: 311-324.

Dosovitskiy A, Ros G, Codevilla F, et al. CARLA: an open urban driving simulator[C]//Conference on Robot Learning. New York: PMLR, 2017: 1-16.

Shah S, Dey D, Lovett C, et al. AirSim: high-fidelity visual and physical simulation for autonomous vehicles[M]//Field and Service Robotics. Cham: Springer International Publishing, 2017: 621-635.

Li Z H, Xia L H, Tang J B, et al. UrbanGPT: spatio-temporal large language models[C]//Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. New York: ACM, 2024: 5351-5362.

Wang J C, Zhang Y, Hu Y L, et al. Large-scale traffic prediction with hierarchical hypergraph message passing networks[J]. IEEE Transactions on Computational Social Systems, 2024, 11(6): 7103-7113.

Yuan Y, Ding J T, Feng J, et al. UniST: a prompt-empowered universal model for urban spatio-temporal prediction[C]//Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. New York: ACM, 2024: 4095-4106.

Liu C X, Yang S, Xu Q X, et al. Spatial-temporal large language model for traffic prediction[C]//Proceedings of the 2024 25th IEEE International Conference on Mobile Data Management (MDM). Piscataway: IEEE Press, 2024: 31-40.

Zhang S Y, Fu D C, Liang W Z, et al. TrafficGPT: Viewing, processing and interacting with traffic foundation models[J]. Transport Policy, 2024, 150: 95-105.

Ke J L, Hu Q Y, Yuan S H, et al. SignalLLM: a general-purpose LLM agent framework for automated signal processing[PP]. V2. (2025-10-30)[2025-12-16]. arXiv: arXiv.2509.17197.

Da L C, Gao M Q, Mei H, et al. Prompt to transfer: sim-to-real transfer for traffic signal control with prompt learning[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2024, 38(1): 82-90.

Hu J J, Dai X Y, Li X J, et al. TrafficWise: leveraging world models for generalized and interpretable traffic control[J]. IEEE Intelligent Transportation Systems Magazine, 2025, 17(5): 50-60.

Zheng O, Abdel-Aty M, Wang D D, et al. TrafficSafetyGPT: tuning a pre-trained large language model to a domain-specific expert in transportation safety[PP]. V1. (2023-07-28)[2025-12-16]. arXiv: arXiv.2307.15311.

Wang L N, Ren Y L, Jiang H, et al. AccidentGPT: a V2X environmental perception multi-modal large model for accident analysis and prevention[C]//Proceedings of the 2024 IEEE Intelligent Vehicles Symposium (Ⅳ). Piscataway: IEEE Press, 2024: 472-477.

Artega C, Park J. A large language model framework to uncover underreporting in traffic crashes[J]. Journal of Safety Research, 2025, 92: 1-13.

Wang X F, Zhu Z, Huang G, et al. DriveDreamer: towards real-world-drive world models forAutonomous driving[C]//Computer Vision - ECCV 2024. Cham: Springer, 2025: 55-72.

Jin Y, Yang R X, Yi Z J, et al. SurrealDriver: designing LLM-powered generative driver agent framework based on human drivers' driving-thinking data[C]//Proceedings of the 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). Piscataway: IEEE Press, 2024: 966-971.

Xu Z H, Zhang Y J, Xie E Z, et al. DriveGPT4: interpretable end-to-end autonomous driving via large language model[J]. IEEE Robotics and Automation Letters, 2024, 9(10): 8186-8193.

Mao J G, Qian Y X, Ye J J, et al. GPT-Driver: learning to drive with GPT[PP]. V3. (2023-12-05)[2025-12-16]. arXiv: arXiv.2310.01415.

Wang Y Q, He J W, Fan L, et al. Driving into the future: multiview visual forecasting and planning with world model for autonomous driving[C]//Proceedings of the 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway: IEEE Press, 2024: 14749-14759.

Sima C, Renz K, Chitta K, et al. DriveLM: driving withGraph visual question answering[C]//Computer Vision - ECCV 2024. Cham: Springer, 2025: 256-274.

Li J J, Qin R, Wang F Y. The future of management: DAO to smart organizations and intelligent operations[J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2023, 53(6): 3389-3399.

Ding W W, Hou J C, Li J J, et al. DeSci based on Web3 and DAO: a comprehensive overview and reference model[J]. IEEE Transactions on Computational Social Systems, 2022, 9(5): 1563-1573.

Qin R, Ding W W, Li J J, et al. Web3-based decentralized autonomous organizations and operations: architectures, models, and mechanisms[J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2023, 53(4): 2073-2082.

王飞跃, 缪青海. 平行科学: 大模型时代AI4S的前沿技术与框架体系[J]. 学术前沿, 2024(14): 64-79.

Wang F Y, Miao Q H. Parallel science: cutting-edge technologies and framework system of AI4S in the era of large models[J]. Frontiers, 2024(14): 64-79.

王飞跃, 王雨桐. 数字科学家与平行科学: AI4S和S4AI的本源与目标[J]. 中国科学院院刊, 2024, 39(1): 27-33.

Wang F Y, Wang Y T. Digital scientists and parallel sciences: the origin and goal of AI for Science and Science for AI[J]. Bulletin of Chinese Academy of Sciences, 2024, 39(1): 27-33.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

平行教学实验室：赋能K-12教育的新AI基础设施

平行澳门野外观测站：全域智能观测与应急管理自主系统

平行奥森：智慧生态与健康生活共融的元宇宙公园

智能的本质与意义

Cerberus：基于深度学习的跨站社交网络机器人检测系统