pith. sign in

arxiv: 2411.14411 · v2 · submitted 2024-11-21 · 💻 cs.LG · cs.MA

Multi-Agent Environments for Vehicle Routing Problems

Ricardo Gama , Ricardo Cunha , Daniel Fuertes , Carlos R. del-Blanco , Hugo L. Fernandes This is my paper

Pith reviewed 2026-05-23 17:10 UTC · model grok-4.3

classification 💻 cs.LG cs.MA
keywords multi-agent reinforcement learningvehicle routing problemsopen source libraryPyTorchreinforcement learning environmentsdiscrete optimizationAEC model
0
0 comments X

The pith

A new PyTorch library supplies a unified modular framework for multi-agent vehicle routing environments across classical, dynamic, stochastic and multi-task variants.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents MAEnvs4VRP as a single library that handles multiple vehicle routing problem types under a multi-agent setup for reinforcement learning. It is built to be flexible so users can customize environments and add new problem types without starting from scratch. The design follows the Agent Environment Cycle model and offers a straightforward API for quick use inside existing RL codebases. The stated goal is to remove the barrier created by scattered or missing open tools, allowing faster algorithm tests and clearer result comparisons between different methods.

Core claim

The authors introduce MAEnvs4VRP, a PyTorch-based library that supplies multi-agent environments for vehicle routing problems. The library supports classical, dynamic, stochastic, and multi-task variants inside one modular architecture that follows the AEC games model and exposes an intuitive API for integration with reinforcement learning frameworks.

What carries the argument

The MAEnvs4VRP library and its modular PyTorch architecture that implements the AEC model to let users switch problem variants and add new ones through a common interface.

If this is right

  • Users can switch between classical, dynamic, stochastic, and multi-task VRP variants inside the same code structure.
  • New routing problems can be added by extending the modular components rather than rewriting environments from scratch.
  • The library can be dropped into existing reinforcement learning training loops with minimal changes because of its AEC compliance and API design.
  • Direct side-by-side evaluation of algorithms becomes possible because all variants share the same interface and observation format.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Standardized environments of this kind could make it easier for OR researchers to import RL methods without building custom simulators first.
  • The same modular pattern might be copied for other discrete optimization tasks that currently lack shared multi-agent test beds.
  • Wider use could surface common failure modes across VRP variants that isolated implementations tend to hide.

Load-bearing premise

The claim that the scarcity of open-source multi-agent VRP frameworks is the main obstacle to testing algorithms and comparing results across studies.

What would settle it

A controlled experiment in which researchers successfully test and compare multiple RL algorithms for several VRP variants using only pre-existing scattered codebases without measurable extra effort or loss of reproducibility.

Figures

Figures reproduced from arXiv: 2411.14411 by Carlos R. del-Blanco, Daniel Fuertes, Hugo L. Fernandes, Ricardo Cunha, Ricardo Gama.

Figure 2.1
Figure 2.1. Figure 2.1: Illustration of a multi-agent VRP instance with four vehicles (left) and the corresponding timeline (right). [PITH_FULL_IMAGE:figures/full_fig_p004_2_1.png] view at source ↗
Figure 3.1
Figure 3.1. Figure 3.1: Schematic representation of the library architecture. [PITH_FULL_IMAGE:figures/full_fig_p005_3_1.png] view at source ↗
Figure 4.1
Figure 4.1. Figure 4.1: Evolution of models performance during training, using single and smallest time agent selection strategies, [PITH_FULL_IMAGE:figures/full_fig_p009_4_1.png] view at source ↗
read the original abstract

Research on Reinforcement Learning (RL) approaches for discrete optimization problems has increased considerably, extending RL to areas classically dominated by Operations Research (OR). Vehicle routing problems are a good example of discrete optimization problems with high practical relevance, for which RL techniques have achieved notable success. Despite these advances, open-source development frameworks remain scarce, hindering both algorithm testing and objective comparison of results. This situation ultimately slows down progress in the field and limits the exchange of ideas between the RL and OR communities. Here, we propose MAEnvs4VRP library, a unified framework for multi-agent vehicle routing environments that supports classical, dynamic, stochastic, and multi-task problem variants within a single modular design. The library, built on PyTorch, provides a flexible and modular architecture design that facilitates customization and the incorporation of new routing problems. It follows the Agent Environment Cycle ("AEC") games model and features an intuitive API, enabling rapid adoption and seamless integration into existing reinforcement learning frameworks. The project source code can be found at https://github.com/ricgama/maenvs4vrp.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript proposes MAEnvs4VRP, a unified open-source library for multi-agent environments targeting vehicle routing problems. It claims to support classical, dynamic, stochastic, and multi-task VRP variants within a single modular PyTorch-based design that follows the Agent Environment Cycle (AEC) model and offers an intuitive API for easy integration with existing RL frameworks. The source code is linked via GitHub.

Significance. If the modular architecture demonstrably implements the claimed variants and enables seamless customization, the library could standardize environments for RL-based VRP research, improving reproducibility and cross-algorithm comparisons between the RL and OR communities. The explicit provision of source code on GitHub is a strength that supports adoption and further development.

major comments (2)
  1. [Abstract] Abstract: the central motivation asserts that 'open-source development frameworks remain scarce' without any citations, references to prior libraries, or feature-comparison table. This premise is load-bearing for positioning MAEnvs4VRP as a 'unified framework'; absent a survey in §1 or §2, the novelty claim cannot be evaluated and risks being incremental rather than enabling.
  2. [Abstract] Abstract: the manuscript describes support for classical/dynamic/stochastic/multi-task variants and a modular design but supplies no validation experiments, benchmark results, or usage examples demonstrating that the AEC-compliant implementation actually works across these variants. This absence undermines assessment of whether the claimed flexibility is realized.
minor comments (1)
  1. The description of the 'intuitive API' and AEC integration would benefit from a short concrete code snippet or pseudocode example to illustrate usage for readers new to the AEC model.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on our manuscript. Below we provide point-by-point responses to the major comments and indicate the changes we will make in the revised version.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central motivation asserts that 'open-source development frameworks remain scarce' without any citations, references to prior libraries, or feature-comparison table. This premise is load-bearing for positioning MAEnvs4VRP as a 'unified framework'; absent a survey in §1 or §2, the novelty claim cannot be evaluated and risks being incremental rather than enabling.

    Authors: We agree that the abstract's motivation would be strengthened by explicit citations and a comparison to prior work. In the revised manuscript we will expand the introduction (Section 1) with a concise survey of existing open-source VRP and RL environment libraries, supported by references, and include a feature-comparison table that positions MAEnvs4VRP relative to them. This will allow readers to evaluate the novelty of the unified multi-agent AEC design. revision: yes

  2. Referee: [Abstract] Abstract: the manuscript describes support for classical/dynamic/stochastic/multi-task variants and a modular design but supplies no validation experiments, benchmark results, or usage examples demonstrating that the AEC-compliant implementation actually works across these variants. This absence undermines assessment of whether the claimed flexibility is realized.

    Authors: The manuscript's primary focus is the library architecture and API rather than algorithmic benchmarking. The accompanying GitHub repository already contains usage examples and tests. To address the concern directly in the paper, we will add a dedicated section presenting concrete usage examples and minimal validation runs that exercise the supported variants (classical, dynamic, stochastic, multi-task) under the AEC model, thereby demonstrating that the claimed modularity and compliance are realized in practice. Full-scale benchmark comparisons with existing solvers remain outside the current scope but can be noted as future work. revision: partial

Circularity Check

0 steps flagged

No circularity; software library proposal with no derivations or fitted quantities

full rationale

The paper introduces the MAEnvs4VRP library as a modular PyTorch-based framework for multi-agent VRP variants following the AEC model. No equations, predictions, parameters, or derivation chains exist that could reduce to inputs by construction. The listed patterns (self-definitional, fitted-input-as-prediction, self-citation load-bearing, etc.) do not apply because there are no mathematical claims or results to inspect for circularity. The motivation statement on scarcity of prior frameworks is a factual premise (potentially debatable via external survey) but does not create any self-referential reduction in a derivation. This matches the default expectation of a non-circular tools paper.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The paper is a software library contribution and contains no mathematical derivations, fitted constants, background axioms, or postulated entities.

pith-pipeline@v0.9.0 · 5723 in / 1160 out tokens · 58199 ms · 2026-05-23T17:10:05.170977+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

56 extracted references · 56 canonical work pages · 2 internal anchors

  1. [1]

    Accorsi, L., Lodi, A., and Vigo, D. (2022). Guidelines for the computational testing of machine learning approaches to vehicle routing problems. Operations Research Letters , 50(2):229--234

    work page 2022

  2. [2]

    V., Christianos, F., and Sch\"afer, L

    Albrecht, S. V., Christianos, F., and Sch\"afer, L. (2024). Multi-Agent Reinforcement Learning: Foundations and Modern Approaches . MIT Press

    work page 2024

  3. [3]

    and Krishnan, K

    Arishi, A. and Krishnan, K. (2023). A multi-agent deep reinforcement learning approach for solving the multi-depot vehicle routing problem. Journal of Management Analytics , 10(3):493--515

    work page 2023

  4. [4]

    M., Jain, A., Luo, R., Maggiar, A., Narayanaswamy, B., and Ye, C

    Balaji, B., Bell-Masterson, J., Bilgin, E., Damianou, A., Garcia, P. M., Jain, A., Luo, R., Maggiar, A., Narayanaswamy, B., and Ye, C. (2019). Orl: Reinforcement learning benchmarks for online stochastic optimization problems. arXiv preprint arXiv:1911.10641

    work page arXiv 2019

  5. [5]

    V., Norouzi, M., and Bengio, S

    Bello, I., Pham, H., Le, Q. V., Norouzi, M., and Bengio, S. (2017). Neural Combinatorial Optimization with Reinforcement Learning . Proceedings of the 5th International Conference on Learning Representations (ICLR)

    work page 2017

  6. [6]

    Berto, F., Hua, C., Luttmann, L., Son, J., Park, J., Ahn, K., Kwon, C., Xie, L., and Park, J. (2024a). Parco: Learning parallel autoregressive policies for efficient multi-agent combinatorial optimization. arXiv preprint arXiv:2409.03811

    work page arXiv

  7. [7]

    Berto, F., Hua, C., Park, J., Kim, M., Kim, H., Son, J., Kim, H., Kim, J., and Park, J. (2023). RL4CO : an extensive reinforcement learning for combinatorial optimization benchmark. arXiv preprint arXiv:2306.17100

    work page arXiv 2023

  8. [8]

    G., Hottung, A., Wouda, N., Lan, L., Tierney, K., and Park, J

    Berto, F., Hua, C., Zepeda, N. G., Hottung, A., Wouda, N., Lan, L., Tierney, K., and Park, J. (2024b). Routefinder: Towards foundation models for vehicle routing problems. arXiv preprint arXiv:2406.15007

    work page arXiv

  9. [9]

    Bettini, M., Prorok, A., and Moens, V. (2024). Benchmarl: Benchmarking multi-agent reinforcement learning. Journal of Machine Learning Research , 25(217):1--10

    work page 2024

  10. [10]

    E., Clark, S., Duplyakin, D., Law, J., and John, P

    Biagioni, D., Tripp, C. E., Clark, S., Duplyakin, D., Law, J., and John, P. C. S. (2022). graphenv: a python library for reinforcement learning on graph search spaces. Journal of Open Source Software , 7(77):4621

    work page 2022

  11. [11]

    Bianchessi, N., Drexl, M., and Irnich, S. (2019). The split delivery vehicle routing problem with time windows and customer inconvenience constraints. Transportation Science , 53(4):1067--1084

    work page 2019

  12. [12]

    I., Kalloniatis, T., Abramowitz, S., Waters, C

    Bonnet, C., Luo, D., Byrne, D., Surana, S., Coyette, V., Duckworth, P., Midgley, L. I., Kalloniatis, T., Abramowitz, S., Waters, C. N., Smit, A. P., Grinsztajn, N., Sob, U. A. M., Mahjoub, O., Tegegn, E., Mimouni, M. A., Boige, R., de Kock, R., Furelos-Blanco, D., Le, V., Pretorius, A., and Laterre, A. (2023). Jumanji: a diverse suite of scalable reinforc...

    work page 2023

  13. [13]

    S., Simonin, O., Matignon, L., and Pereyron, F

    Bono, G., Dibangoye, J. S., Simonin, O., Matignon, L., and Pereyron, F. (2020). Solving multi-agent routing problems using deep attention mechanisms. IEEE Transactions on Intelligent Transportation Systems , 22(12):7804--7813

    work page 2020

  14. [14]

    D., and Moens, V

    Bou, A., Bettini, M., Dittert, S., Kumar, V., Sodhani, S., Yang, X., Fabritiis, G. D., and Moens, V. (2023). Torchrl: A data-driven decision-making library for pytorch

    work page 2023

  15. [15]

    Braekers, K., Ramaekers, K., and Van Nieuwenhuyse, I. (2016). The vehicle routing problem: State of the art classification and review. Computers & industrial engineering , 99:300--313

    work page 2016

  16. [16]

    Brockman, G., Cheung, V., Pettersson, L., Schneider, J., Schulman, J., Tang, J., and Zaremba, W. (2016). Openai gym. arXiv preprint arXiv:1606.01540

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  17. [17]

    Dumas, Y., Desrosiers, J., and Soumis, F. (1991). The pickup and delivery problem with time windows. European Journal of Operational Research , 54(1):7--22

    work page 1991

  18. [18]

    Figliozzi, M. A. (2010). An iterative route construction and improvement algorithm for the vehicle routing problem with soft time windows. Transportation Research Part C: Emerging Technologies , 18(5):668--679. Applications of Advanced Technologies in Transportation: Selected papers from the 10th AATT Conference

    work page 2010

  19. [19]

    R., Jaureguizar, F., and Garc \' a, N

    Fuertes, D., del Blanco, C. R., Jaureguizar, F., and Garc \' a, N. (2023). Solving the team orienteering problem with transformers. arXiv preprint arXiv:2311.18662

    work page arXiv 2023

  20. [20]

    Guo, F., Wei, Q., Wang, M., Guo, Z., and Wallace, S. W. (2023). Deep attention models with dimension-reduction and gate mechanisms for solving practical time-dependent vehicle routing problems. Transportation Research Part E: Logistics and Transportation Review , 173:103095

    work page 2023

  21. [21]

    Hu, S., Zhong, Y., Gao, M., Wang, W., Dong, H., Liang, X., Li, Z., Chang, X., and Yang, Y. (2023). Marllib: A scalable and efficient multi-agent reinforcement learning library. Journal of Machine Learning Research , 24(315):1--23

    work page 2023

  22. [22]

    D., Perez, H

    Hubbs, C. D., Perez, H. D., Sarwar, O., Sahinidis, N. V., Grossmann, I. E., and Wassick, J. M. (2020). Or-gym: A reinforcement learning library for operations research problems

    work page 2020

  23. [23]

    Kim, M., Park, J., and Park, J. (2022). Sym-nco: Leveraging symmetricity for neural combinatorial optimization. Advances in Neural Information Processing Systems , 35:1936--1949

    work page 2022

  24. [24]

    Kingma, D. P. and Ba, J. (2015). Adam: A method for stochastic optimization. International Conference on Learning Representations (ICLR) , abs/1412.6980

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  25. [25]

    Kool, W., Van Hoof , H., and Welling, M. (2019). Attention, learn to solve routing problems! 7th International Conference on Learning Representations, ICLR 2019 , pages 1--25

    work page 2019

  26. [26]

    Kwon, Y.-D., Choo, J., Kim, B., Yoon, I., Gwon, Y., and Min, S. (2020). Pomo: Policy optimization with multiple optima for reinforcement learning. Advances in Neural Information Processing Systems , 33:21188--21198

    work page 2020

  27. [27]

    Li, B., Wu, G., He, Y., Fan, M., and Pedrycz, W. (2022). An overview and experimental study of learning-based optimization algorithms for the vehicle routing problem. IEEE/CAA Journal of Automatica Sinica , 9(7):1115--1138

    work page 2022

  28. [28]

    Li, J., Niu, Y., Zhu, G., and Xiao, J. (2024). Solving pick-up and delivery problems via deep reinforcement learning based symmetric neural optimization. Expert Systems with Applications , 255:124514

    work page 2024

  29. [29]

    Liu, F., Lin, X., Zhang, Q., Tong, X., and Yuan, M. (2024a). Multi-task learning for routing problem with cross-problem zero-shot generalization. arXiv preprint arXiv:2402.16891

    work page arXiv

  30. [30]

    Liu, Q., Liu, C., Niu, S., Long, C., Zhang, J., and Xu, M. (2024b). 2d-ptr: 2d array pointer network for solving the heterogeneous capacitated vehicle routing problem. In Proceedings of the 23rd International Conference on Autonomous Agents and Multiagent Systems , pages 1238--1246

    work page

  31. [31]

    Mazyavkina, N., Sviridov, S., Ivanov, S., and Burnaev, E. (2021). Reinforcement learning for combinatorial optimization: A survey. Computers & Operations Research , 134:105400

    work page 2021

  32. [32]

    W., Tracey, B

    Menda, K., Chen, Y.-C., Grana, J., Bono, J. W., Tracey, B. D., Kochenderfer, M. J., and Wolpert, D. (2018). Deep reinforcement learning for event-driven multi-agent decision processes. IEEE Transactions on Intelligent Transportation Systems , 20(4):1259--1268

    work page 2018

  33. [33]

    P., Nygren, E., Laurent, F., Schneider, M., Scheller, C

    Mohanty, S. P., Nygren, E., Laurent, F., Schneider, M., Scheller, C. V., Bhattacharya, N., Watson, J. D., Egli, A., Eichenberger, C., Baumberger, C., Vienken, G., Sturm, I., Sartoretti, G., and Spigler, G. (2020). Flatland-rl : Multi-agent reinforcement learning on trains. ArXiv , abs/2012.05893

    work page arXiv 2020

  34. [34]

    V., and Tak \' a c , M

    Nazari, M., Oroojlooy, A., Snyder, L. V., and Tak \' a c , M. (2018). Deep Reinforcement Learning for Solving the Vehicle Routing Problem . In Proceedings Neural Information Processing Systems (NIPS) , pages 9839--9849

    work page 2018

  35. [35]

    and Liu, S

    Pan, W. and Liu, S. Q. (2023). Deep reinforcement learning for the dynamic and uncertain vehicle routing problem. Applied Intelligence , 53(1):405--422

    work page 2023

  36. [36]

    Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., Desmaison, A., K\" o pf, A., Yang, E., DeVito, Z., Raison, M., Tejani, A., Chilamkurthy, S., Steiner, B., Fang, L., Bai, J., and Chintala, S. (2019). PyTorch: an imperative style, high-performance deep learning library . Curran Associat...

    work page 2019

  37. [37]

    Raffin, A., Hill, A., Gleave, A., Kanervisto, A., Ernestus, M., and Dormann, N. (2021). Stable-baselines3: Reliable reinforcement learning implementations. Journal of Machine Learning Research , 22(268):1--8

    work page 2021

  38. [38]

    and Niu, L

    Shi, R. and Niu, L. (2023). A brief survey on learning based methods for vehicle routing problems. Procedia Computer Science , 221:773--780. Tenth International Conference on Information Technology and Quantitative Management (ITQM 2023)

    work page 2023

  39. [39]

    and Leyton-Brown, K

    Shoham, Y. and Leyton-Brown, K. (2008). Multiagent systems: Algorithmic, game-theoretic, and logical foundations . Cambridge University Press

    work page 2008

  40. [40]

    Solomon, M. M. (1987). Algorithms for the vehicle routing and scheduling problems with time window constraints. Operations Research , 35(2):254--265

    work page 1987

  41. [41]

    S., Dieffendahl, C., Horsch, C., Perez-Vicente, R., Williams, N., Lokesh, Y., and Ravi, P

    Terry, J., Black, B., Grammel, N., Jayakumar, M., Hari, A., Sullivan, R., Santos, L. S., Dieffendahl, C., Horsch, C., Perez-Vicente, R., Williams, N., Lokesh, Y., and Ravi, P. (2021). Pettingzoo: Gym for multi-agent reinforcement learning. In Ranzato, M., Beygelzimer, A., Dauphin, Y., Liang, P., and Vaughan, J. W., editors, Advances in Neural Information ...

    work page 2021

  42. [42]

    K., and Schmidt-Thieme, L

    Thyssens, D., Dernedde, T., Falkner, J. K., and Schmidt-Thieme, L. (2023). Routing arena: A benchmark suite for neural routing solvers. arXiv preprint arXiv:2310.04140

    work page arXiv 2023

  43. [43]

    and Gunawan, A

    Vansteenwegen, P. and Gunawan, A. (2019). Orienteering Problems, Models and Algorithms for Vehicle Routing Problems with Profits . Springer, euro advan edition

    work page 2019

  44. [44]

    Vaswani, A. (2017). Attention is all you need. Advances in Neural Information Processing Systems

    work page 2017

  45. [45]

    Vinyals, O., Fortunato, M., and Jaitly, N. (2015). Pointer networks. In Cortes, C., Lawrence, N. D., Lee, D. D., Sugiyama, M., and Garnett, R., editors, Advances in Neural Information Processing Systems 28 , pages 2692--2700. Curran Associates, Inc

    work page 2015

  46. [46]

    P., Li, T., and Wang, J

    Wan, C. P., Li, T., and Wang, J. M. (2023). Rlor: A flexible framework of deep reinforcement learning for operation research

    work page 2023

  47. [47]

    Williams, R. J. (1992). Simple statistical gradient-following algorithms for connectionist reinforcement learning . Machine Learning , 8(3):229--256

    work page 1992

  48. [48]

    A., Lan, L., and Kool, W

    Wouda, N. A., Lan, L., and Kool, W. (2024). PyVRP : a high-performance VRP solver package. INFORMS Journal on Computing

    work page 2024

  49. [49]

    L., and Zhou, Y

    Wu, X., Wang, D., Wen, L., Xiao, Y., Wu, C., Wu, Y., Yu, C., Maskell, D. L., and Zhou, Y. (2024). Neural combinatorial optimization algorithms for solving vehicle routing problems: A comprehensive survey with perspectives. arXiv preprint arXiv:2406.00415

    work page arXiv 2024

  50. [50]

    Xiang, C., Wu, Z., Tu, J., and Huang, J. (2024). Centralized deep reinforcement learning method for dynamic multi-vehicle pickup and delivery problem with crowdshippers. IEEE Transactions on Intelligent Transportation Systems , 25(8):9253--9267

    work page 2024

  51. [51]

    Zhang, K., He, F., Zhang, Z., Lin, X., and Li, M. (2020). Multi-vehicle routing problems with soft time windows: A multi-agent reinforcement learning approach. Transportation Research Part C: Emerging Technologies , 121:102861

    work page 2020

  52. [52]

    R., Reijnen, R., Catshoek, T., Vos, D., Verwer, S., Schmitt-Ulms, F., Hottung, A., et al

    Zhang, Y., Bliek, L., da Costa, P., Afshar, R. R., Reijnen, R., Catshoek, T., Vos, D., Verwer, S., Schmitt-Ulms, F., Hottung, A., et al. (2023a). The first ai4tsp competition: Learning to solve stochastic routing problems. Artificial Intelligence , 319:103918

    work page

  53. [53]

    Zhang, Z., Qi, G., and Guan, W. (2023b). Coordinated multi-agent hierarchical deep reinforcement learning to solve multi-trip vehicle routing problems with soft time windows. IET Intelligent Transport Systems , 17(10):2034--2051

    work page 2034

  54. [54]

    Zhou, G., Li, X., Li, D., and Bian, J. (2024a). Learning-based optimization algorithms for routing problems: Bibliometric analysis and literature review. IEEE Transactions on Intelligent Transportation Systems

    work page

  55. [55]

    Zhou, J., Cao, Z., Wu, Y., Song, W., Ma, Y., Zhang, J., and Xu, C. (2024b). Mvmoe: Multi-task vehicle routing solver with mixture-of-experts. arXiv preprint arXiv:2405.01029

    work page arXiv

  56. [56]

    Zong, Z., Zheng, M., Li, Y., and Jin, D. (2022). Mapdp: Cooperative multi-agent reinforcement learning to solve pickup and delivery problems. In Proceedings of the AAAI Conference on Artificial Intelligence , volume 36, pages 9980--9988

    work page 2022