Skip to main content
SpringerLink
Log in

Open-ti: open traffic intelligence with augmented language model

  • Original Article
  • Published:
International Journal of Machine Learning and Cybernetics Aims and scope Submit manuscript

Abstract

Transportation has greatly benefited the cities’ development in the modern civilization process. Intelligent transportation, leveraging advanced computer algorithms, could further increase people’s daily commuting efficiency. However, intelligent transportation, as a cross-discipline, often requires practitioners to comprehend complicated algorithms and obscure neural networks, bringing a challenge for the advanced techniques to be trusted and deployed in practical industries. Recognizing the expressiveness of the pre-trained large language models, especially the potential of being augmented with abilities to understand and execute intricate commands, we introduce Open-TI. Serving as a bridge to mitigate the industry-academic gap, Open-TI is an innovative model targeting the goal of Turing Indistinguishable Traffic Intelligence, it is augmented with the capability to harness external traffic analysis packages based on existing conversations. Marking its distinction, Open-TI is the first method capable of conducting exhaustive traffic analysis from scratch—spanning from map data acquisition to the eventual execution in complex simulations. Besides, Open-TI is able to conduct task-specific embodiment like training and adapting the traffic signal control policies (TSC), explore demand optimizations, etc. Furthermore, we explored the viability of LLMs directly serving as control agents, by understanding the expected intentions from Open-TI, we designed an agent-to-agent communication mode to support Open-TI conveying messages to ChatZero (control agent), and then the control agent would choose from the action space to proceed the execution. We eventually provide the formal implementation structure, and the open-ended design invites further community-driven enhancements. A demo video is provided at: https://youtu.be/pZ4-5PXz9Xs.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Algorithm 1
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13

Similar content being viewed by others

Data availability

All datasets and road networks are publicly available at: https://github.com/DaRL-LibSignal/LibSignalhttps://traffic-signal-control.github.io/dataset.html.

Notes

  1. We have released the code at: https://github.com/DaRL-LibSignal/OpenTI.git.

  2. https://blog.google/products/search/search-language-understanding-bert/.

  3. https://traffic-signal-control.github.io/dataset.html.

References

  1. Yukawa S, Kikuchi M (1995) Coupled-map modeling of one-dimensional traffic flow. J Phys Soc Jpn 64(1):35–38

    Article  Google Scholar 

  2. Chao Q, Bi H, Li W, Mao T, Wang Z, Lin MC, Deng Z (2020) A survey on visual traffic simulation: Models, evaluations, and applications in autonomous driving. In: Computer Graphics Forum, vol. 39, pp. 287–308. Wiley Online Library

  3. Dai Z, Liu XC, Chen X, Ma X (2020) Joint optimization of scheduling and capacity for mixed traffic with autonomous and human-driven buses: A dynamic programming approach. Transportation Research Part C: Emerging Technologies 114:598–619

    Article  Google Scholar 

  4. Zhou XS, Cheng Q, Wu X, Li P, Belezamo B, Lu J, Abbasi M (2022) A meso-to-macro cross-resolution performance approach for connecting polynomial arrival queue model to volume-delay function with inflow demand-to-capacity ratio. Multimodal Transp 1(2):100017

    Article  Google Scholar 

  5. Wei H, Xu N, Zhang H, Zheng G, Zang X, Chen C, Zhang W, Zhu Y, Xu K, Li Z (2019) Colight: Learning network-level cooperation for traffic signal control. In: Proceedings of the 28th ACM International Conference on Information and Knowledge Management, pp. 1913–1922

  6. Osorio C (2019) High-dimensional offline origin-destination (od) demand calibration for stochastic traffic simulators of large-scale road networks. Transportation Research Part B: Methodological 124:18–43

    Article  Google Scholar 

  7. Lopez PA, Behrisch M, Bieker-Walz L, Erdmann J, Flötteröd Y-P, Hilbrich R, Lücken L, Rummel J, Wagner P, Wießner E (2018) Microscopic traffic simulation using sumo. In: 2018 21st International Conference on Intelligent Transportation Systems (ITSC), pp. 2575–2582. IEEE

  8. Zhang H, Feng S, Liu C, Ding Y, Zhu Y, Zhou Z, Zhang W, Yu Y, Jin H, Li Z (2019) Cityflow: A multi-agent reinforcement learning environment for large scale city traffic scenario. In: The World Wide Web Conference, pp. 3620–3624

  9. Fellendorf M, Vortisch P (2010) Microscopic traffic flow simulator vissim. Fundamentals of traffic simulation, 63–93

  10. Lu J, Zhou XS (2023) Virtual track networks: A hierarchical modeling framework and open-source tools for simplified and efficient connected and automated mobility (cam) system design based on general modeling network specification (gmns). Transportation Research Part C: Emerging Technologies 153:104223

    Article  Google Scholar 

  11. Zhang S, Fu D, Zhang Z, Yu B, Cai P (2023) Trafficgpt: Viewing, processing and interacting with traffic foundation models. arXiv preprint arXiv:2309.06719

  12. de Zarzà I, de Curtò J, Roig G, Calafate CT (2023) Llm multimodal traffic accident forecasting. Sensors 23(22):9225

    Article  Google Scholar 

  13. Vaithilingam P, Zhang T, Glassman EL (2022) Expectation vs. experience: Evaluating the usability of code generation tools powered by large language models. In: Chi Conference on Human Factors in Computing Systems Extended Abstracts, pp. 1–7

  14. Li Y, Gao C, Song X, Wang X, Xu Y, Han S (2023) Druggpt: A gpt-based strategy for designing potential ligands targeting specific proteins. bioRxiv, 2023–06

  15. Tang J, Yang Y, Wei W, Shi L, Su L, Cheng S, Yin D, Huang C (2023) Graphgpt: Graph instruction tuning for large language models. arXiv preprint arXiv:2310.13023

  16. Mialon G, Dessì R, Lomeli M, Nalmpantis C, Pasunuru R, Raileanu R, Rozière B, Schick T, Dwivedi-Yu J, Celikyilmaz A, et al. (2023) Augmented language models: a survey. arXiv preprint arXiv:2302.07842

  17. Devlin J, Chang M-W, Lee K, Toutanova K (2018) Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805

  18. Brown T, Mann B, Ryder N, Subbiah M, Kaplan JD, Dhariwal P, Neelakantan A, Shyam P, Sastry G, Askell A et al (2020) Language models are few-shot learners. Adv Neural Inf Process Syst 33:1877–1901

    Google Scholar 

  19. Chowdhery A, Narang S, Devlin J, Bosma M, Mishra G, Roberts A, Barham P, Chung HW, Sutton C, Gehrmann S et al (2023) Palm: Scaling language modeling with pathways. J Mach Learn Res 24(240):1–113

    Google Scholar 

  20. Chen M, Tworek J, Jun H, Yuan Q, Pinto HPdO, Kaplan J, Edwards H, Burda Y, Joseph N, Brockman G et al. (2021) Evaluating large language models trained on code. arXiv preprint arXiv:2107.03374

  21. Liu Y, Han T, Ma S, Zhang J, Yang Y, Tian J, He H, Li A, He M, Liu Z, et al. (2023) Summary of chatgpt-related research and perspective towards the future of large language models. Meta-Radiology, 100017

  22. Tirumala K, Markosyan A, Zettlemoyer L, Aghajanyan A (2022) Memorization without overfitting: Analyzing the training dynamics of large language models. Adv Neural Inf Process Syst 35:38274–38290

    Google Scholar 

  23. Zhou D, Schärli N, Hou L, Wei J, Scales N, Wang X, Schuurmans D, Cui C, Bousquet O, Le Q et al. (2022) Least-to-most prompting enables complex reasoning in large language models. arXiv preprint arXiv:2205.10625

  24. Da L, Gao M, Mei H, Wei H. (2023) Llm powered sim-to-real transfer for traffic signal control. arXiv preprint arXiv:2308.14284

  25. Li M, Song F, Yu B, Yu H, Li Z, Huang F, Li Y (2023) Api-bank: A benchmark for tool-augmented llms. arXiv preprint arXiv:2304.08244

  26. Wang Y, Ma X, Chen W (2023) Augmenting black-box llms with medical textbooks for clinical question answering. arXiv preprint arXiv:2309.02233

  27. Liang Y, Wu C, Song T, Wu W, Xia Y, Liu Y, Ou Y, Lu S, Ji L, Mao S et al. (2023) Taskmatrix. ai: Completing tasks by connecting foundation models with millions of apis. arXiv preprint arXiv:2303.16434

  28. Liu C, Yang S, Xu Q, Li Z, Long C, Li Z, Zhao R (2024) Spatial-temporal large language model for traffic prediction. arXiv preprint arXiv:2401.10134

  29. Tong L, Pan Y, Shang P, Guo J, Xian K, Zhou X (2019) Open-source public transportation mobility simulation engine dtalite-s: A discretized space-time network-based modeling framework for bridging multi-agent simulation and optimization. Urban Rail Transit 5:1–16

    Article  Google Scholar 

  30. Dosovitskiy A, Ros G, Codevilla F, Lopez A, Koltun V (2017) Carla: An open urban driving simulator. In: Conference on Robot Learning, pp. 1–16. PMLR

  31. Mullakkal-Babu FA, Wang M, van Arem B, Shyrokau B, Happee R (2020) A hybrid submicroscopic-microscopic traffic flow simulation framework. IEEE Trans Intell Transp Syst 22(6):3430–3443

    Article  Google Scholar 

  32. de Souza F, Verbas O, Auld J (2019) Mesoscopic traffic flow model for agent-based simulation. Procedia Computer Science 151:858–863

    Article  Google Scholar 

  33. Oppe S (1989) Macroscopic models for traffic and traffic safety. Accident Analysis & Prevention 21(3):225–232

    Article  MathSciNet  Google Scholar 

  34. Boukerche A, Tao Y, Sun P (2020) Artificial intelligence-based vehicular traffic flow prediction methods for supporting intelligent transportation systems. Comput Netw 182:107484

    Article  Google Scholar 

  35. Masek P, Masek J, Frantik P, Fujdiak R, Ometov A, Hosek J, Andreev S, Mlynek P, Misurec J (2016) A harmonized perspective on transportation management in smart cities: The novel iot-driven environment for road traffic modeling. Sensors 16(11):1872

    Article  Google Scholar 

  36. Maroto J, Delso E, Felez J, Cabanellas JM (2006) Real-time traffic simulation with a microscopic model. IEEE Trans Intell Transp Syst 7(4):513–527

    Article  Google Scholar 

  37. NVIDIA: Simulation for self-driving vehicles (2023)

  38. Gulino C, Fu J, Luo W, Tucker G, Bronstein E, Lu Y, Harb J, Pan X, Wang Y, Chen X et al. (2023) Waymax: An accelerated, data-driven simulator for large-scale autonomous driving research. arXiv preprint arXiv:2310.08710

  39. Behrisch M, Bieker L, Erdmann J, Krajzewicz D (2011) Sumo–simulation of urban mobility: an overview. In: Proceedings of SIMUL 2011, The Third International Conference on Advances in System Simulation . ThinkMind

  40. Qadri SSSM, Gökçe MA, Öner E (2020) State-of-art review of traffic signal control methods: challenges and opportunities. Eur Transp Res Rev 12:1–23

    Article  Google Scholar 

  41. Wei H, Zheng G, Yao H, Li Z (2018) Intellilight: A reinforcement learning approach for intelligent traffic light control. In: Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pp. 2496–2505

  42. Willumsen LG (1978) Estimation of an od matrix from traffic counts-a review

  43. Abrahamsson T (1998) Estimation of origin-destination matrices using traffic counts-a literature survey

  44. Medina A, Taft N, Salamatian K, Bhattacharyya S, Diot C (2002) Traffic matrix estimation: Existing techniques and new directions. ACM SIGCOMM Computer Communication Review 32(4):161–174

    Article  Google Scholar 

  45. Mahmassani HS (2001) Dynamic network traffic assignment and simulation methodology for advanced system management applications. Netw Spat Econ 1:267–292

    Article  Google Scholar 

  46. Mahmassani HS, Zhou X (2005) In: Abed, E.H. (ed.) Transportation System Intelligence: Performance Measurement and Real-Time Traffic Estimation and Prediction in a Day-to-Day Learning Framework, pp. 305–328. Birkhäuser Boston, Boston, MA

  47. Zhou X, Qin X, Mahmassani HS (2003) Dynamic origin-destination demand estimation with multiday link traffic counts for planning applications. Transp Res Rec 1831(1):30–38

    Article  Google Scholar 

  48. Zhou X, Erdoğan S, Mahmassani HS (2006) Dynamic origin-destination trip demand estimation for subarea analysis. Transp Res Rec 1964(1):176–184

    Article  Google Scholar 

  49. Zhou X, List GF (2010) An information-theoretic sensor location model for traffic origin-destination demand estimation applications. Transp Sci 44(2):254–273

    Article  Google Scholar 

  50. Zhou X, Lu C, Zhang K (2013) Dynamic origin-destination demand flow estimation utilizing heterogeneous data sources under congested traffic conditions

  51. Krishnakumari P, Van Lint H, Djukic T, Cats O (2020) A data driven method for od matrix estimation. Transportation Research Part C: Emerging Technologies 113:38–56

    Article  Google Scholar 

  52. Fedorov A, Nikolskaia K, Ivanov S, Shepelev V, Minbaleev A (2019) Traffic flow estimation with data from a video surveillance camera. Journal of Big Data 6:1–15

    Article  Google Scholar 

  53. Pamuła T, Żochowska R (2023) Estimation and prediction of the od matrix in uncongested urban road network based on traffic flows using deep learning. Eng Appl Artif Intell 117:105550

    Article  Google Scholar 

  54. Fu H, Lam WH, Shao H, Kattan L, Salari M (2022) Optimization of multi-type traffic sensor locations for estimation of multi-period origin-destination demands with covariance effects. Transportation Research Part E: Logistics and Transportation Review 157:102555

    Article  Google Scholar 

  55. Kumarage S, Yildirimoglu M, Zheng Z (2023) A hybrid modelling framework for the estimation of dynamic origin-destination flows. Transportation Research Part B: Methodological 176:102804

    Article  Google Scholar 

  56. Mei H, Lei X, Da L, Shi B, Wei H (2023) Libsignal: an open library for traffic signal control. Machine Learning, 1–37

  57. Cools S-B, Gershenson C, D’Hooghe B (2013) Self-organizing traffic lights: A realistic simulation. Advances in applied self-organizing systems, 45–55

  58. Wei H, Zheng G, Gayah V, Li Z (2021) Recent advances in reinforcement learning for traffic signal control: A survey of models and evaluation. ACM SIGKDD Explorations Newsl 22(2):12–18

    Article  Google Scholar 

  59. Zheng G, Zang X, Xu N, Wei H, Yu Z, Gayah V, Xu K, Li Z (2019) Diagnosing reinforcement learning for traffic signal control. arXiv preprint arXiv:1905.04716

  60. Wei H, Chen C, Zheng G, Wu K, Gayah V, Xu K, Li Z (2019) Presslight: Learning max pressure control to coordinate traffic signals in arterial network. In: Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pp. 1290–1298

Download references

Author information

Authors and Affiliations

  1. School of Computing and Augmented Intelligence, Arizona State University, 350 E Lemon St, Tempe, AZ, 85287, USA

    Longchao Da, Kuanru Liou, Tiejin Chen, Yezhou Yang & Hua Wei

  2. School of Sustainable Engineering and the Built Environment, Arizona State University, 350 E Lemon St, Tempe, AZ, 85287, USA

    Xuesong Zhou & Xiangyong Luo

Authors
  1. Longchao Da
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Kuanru Liou
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Tiejin Chen
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Xuesong Zhou
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Xiangyong Luo
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Yezhou Yang
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Hua Wei
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Hua Wei.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix A

Appendix A

1.1 Detailed explanation of acronyms

See Tables 8, 9.

Table 8 The detailed of Config and corresponding roade netowrk
Full size table
Table 9 A glance at the technique terms and abbreviation
Full size table

1.2 Thought chain process examples

In this section, we provide more Chain-of-thought (CoT) process examples, as a reflection on given a task, how the Open-TI thinks and proposes the solutions, and how it searches in the augmentation tools to further provide analysis.

We have shown the requests such as: downloading OSM files of specific locations, interpreting the log files, showing areas on a map, filtering assigned lane types from a given map, generating demand files based on a map file, executing multiple simulations like DLSim, SUMO, etc., running LibSignal for traffic signal control (Figs. 14, 15, 16, 17, 18, 19, 20, 21, 22, 23).

Fig. 14
figure 14

Ask Open-TI to get the geographic information of Arizona State University, Tempe Campus

Full size image
Fig. 15
figure 15

Ask Open-TI to download OSM data of Arizona State University, Tempe Campus

Full size image
Fig. 16
figure 16

Ask Open-TI to download OSM data of Arizona State University, Tempe Campus without giving Open-TI geographic information

Full size image
Fig. 17
figure 17

Ask Open-TI to analyze the interested log file in a specific path

Full size image
Fig. 18
figure 18

Ask Open-TI to generate demand file through downloaded osm data of Arizona State University, Tempe Campus

Full size image
Fig. 19
figure 19

Ask Open-TI to show the map of Arizona State University, Tempe Campus

Full size image
Fig. 20
figure 20

Ask Open-TI to show the map of Taipei 101 without giving Open-TI geographic information

Full size image
Fig. 21
figure 21

Ask Open-TI to filter the bikable area of Arizona State University, Tempe Campus

Full size image
Fig. 22
figure 22

Ask Open-TI to run the demand file of Arizona State University, Tempe Campus on DLSim

Full size image
Fig. 23
figure 23

Ask Open-TI to run Libsignal on CityFlow environment, DQN policy, and episode 10

Full size image

1.3 Other interactions with Open-TI examples

This section provides more examples of user interactions, including result interpretation, log file analysis, O-D matrix optimization, etc (Figs. 24, 25, 26, 27).

Fig. 24
figure 24

Ask Open-TI to show the map of interested place, download.osm data of interested place, use the OSM file of target place to filter railway routes, and conduct traffic signal control

Full size image
Fig. 25
figure 25

An example of asking Open-TI to generate demand file from OSM file, run simulator on Libsignal in different algorithm and episode, execute simulation on DLSim, understand and response in multi-languages, and analyze logs in a specific path

Full size image
Fig. 26
figure 26

An example of asking Open-TI to analyze file with a specific path. In this case, run the simulation on Libsignal using simulator CityFlow, algorithm fixedtime, and episode 5. Additionally, run the simulation on Libsignal using simulator CityFlow, algorithm DQN, and episode 5. Finally, use the logAnalyzer to compare the performance of both algorithms

Full size image
Fig. 27
figure 27

The demonstration on how Open-TI is used to optimize the OD-Demand matrix. It first visualizes the defined traffic zone information, and sets the observation point to mimic the real-world data collection process. Then based on the gap between simulation observation and real-world observation (count data), the O-D matrix is optimized to mitigate the observation gap by optimization algorithms (e.g., Genetic Algorithm). After the optimization, the final O-D matrix is simulated again, and the comparison of observation is shown in the end

Full size image

1.4 Unexpected use cases and improvement potentials

This section provides the unexpected use cases of Open-TI, including the corner case and multi-step problems (Figs. 28, 29).

Fig. 28
figure 28

The figure illustrates that when the agent is tasked with autoDownloadOpenStreetMapFile for the target location but is provided with incorrect longitude and latitude order, it results in an error being raised and failure to retrieve the data. The geographical information provided does not adhere to the default order of [min_long, min_lat, max_long, max_lat], and does not explicitly specify the geographic meaning of each value. This will require more comprehensive corner case processing and information examination

Full size image
Fig. 29
figure 29

The figure illustrates that we asked the agent to display the map of the target place but did not provide geographical information. It requires two steps: first, calling queryAreaRange and then inputting the result into showOnMap. However, the call to queryAreaRange stops prematurely, leading to an API mismatch. The alignment between intermediate output and connected sub-tasks still needs to be explored in the future

Full size image

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Da, L., Liou, K., Chen, T. et al. Open-ti: open traffic intelligence with augmented language model. Int. J. Mach. Learn. & Cyber. (2024). https://doi.org/10.1007/s13042-024-02190-8

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1007/s13042-024-02190-8

Keywords

Search

Navigation