Hierarchical multiagent reinforcement learning schemes for air traffic management


In this work we investigate the use of hierarchical multiagent reinforcement learning methods for the computation of policies to resolve congestion problems in the air traffic management domain. To address cases where the demand of airspace use exceeds capacity, we consider agents representing flights, who need to decide on ground delays at the pre-tactical stage of operations, towards executing their trajectories while adhering to airspace capacity constraints. Hierarchical reinforcement learning manages to handle real-world problems with high complexity, by partitioning the task into hierarchies of states and/or actions. This provides an efficient way of exploring the state–action space and constructing an advantageous decision-making mechanism. We first establish a general framework of hierarchical multiagent reinforcement learning, and then, we further formulate four alternative schemes of abstractions, on states, actions, or both. To quantitatively assess the quality of solutions of the proposed approaches and show the potential of the hierarchical methods in resolving the demand–capacity balance problem, we provide experimental results on real-world evaluation cases, where we measure the average delay per flight and the number of flights with delays.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2


  1. 1.


  2. 2.

    The mapping of joint states to abstract joint states is straightforward using \(\phi _L\), given that each joint state is a concatenation of local state parameters.


  1. 1.

    Abel D, Hershkowitz DE, Littman ML (2016) Near optimal behavior via approximate state abstraction. In: International conference on machine learning (ICML‘16), vol 48, pp 2915–2923

  2. 2.

    Agogino AK, Tumer K (2012) A multiagent approach to managing air traffic flow. Auton Agents Multiagent Syst 24:1–25

    Article  Google Scholar 

  3. 3.

    Andreas J, Klein D, Levine S (2017) Modular multitask reinforcement learning with policy sketches. In: 34th international conference on machine learning (ICML), pp 166–1751

  4. 4.

    Andrienko G, Andrienko N, Bak P, Keim D, Wrobel S (2013) Visual analytics of movement. Springer, Berlin

    Google Scholar 

  5. 5.

    Bacon P, Harb J, Precup D (2017) The option-critic architecture. In: Proceedings of the thirty-first AAAI conference on artificial intelligence, AAAI’17, pp 1726–1734. AAAI Press

  6. 6.

    Bai A, Srivastava S, Russell S (2016) Markovian state and action abstractions for MDPs via hierarchical MCTS. In: Proceedings of the twenty-fifth international joint conference on artificial intelligence, IJCAI’16, pp 3029–3037. AAAI Press

  7. 7.

    Bazzan ALC, Wahle J, Klügl F (1999) Agents in traffic modelling—from reactive to social behaviour. In: 23rd annual german conference on artificial intelligence, pp 303–306

  8. 8.

    Chen J, Wang Z, Tomizuka M (2018) Deep hierarchical reinforcement learning for autonomous driving with distinct behaviors. In: 2018 IEEE intelligent vehicles symposium (IV), pp 1239–1244

  9. 9.

    Colby M, Tumer K (2013) Multiagent reinforcement learnng in a distributed sensor network with indirect feedback. In: International conference on autonomous agents and multi-agent systems (AAMAS’13), pp 941–948

  10. 10.

    Cook AJ, Tanner G (2015) European airline delay cost reference values. http://www.eurocontrol.int/publications/european-airline-delaycost-reference-values

  11. 11.

    Dayan P, Hinton GE (1992) Feudal reinforcement learning. In: Advances in neural information processing systems, [NIPS Conference], vol 5, pp 271–278

  12. 12.

    Delvin S, Yliniemi L, Kudenko D, Tumer K (2014) Potential-based difference rewars for multiagent reinforcement learning. In: International conference on autonomous agents and multi-agent systems (AAMAS’14), pp 165–172

  13. 13.

    Dietterich T (2000) Hierarchical reinforcement learning with the maxq value function decomposition. J Artif Intell Res 13:227–303

    MathSciNet  Article  Google Scholar 

  14. 14.

    Frans K, Ho J, Chen X, Abbeel P, Schulman J (2017) Meta learning shared hierarchies. Technical Report. arXiv preprint arXiv:1710.09767

  15. 15.

    Guestrin C, Lagoudakis M, Parr R (2002) Coordinated reinforcement learning. In: International conference on machine learning (ICML‘02), pp 227–234

  16. 16.

    Jong N, Stone P (2005) State abstraction discovery from irrelevant state variables. In: International joint conference on artificial intelligence (IJCAI ’05), pp 752–757

  17. 17.

    Karp R, Koutsoupias E, Papadimitriou C, Shenker S (2000) Optimization problems in congestion control. In: 16th Annual symposium on foundations of computer science, pp 66–74

  18. 18.

    Kok JR, Vlassis N (2006) Collaborative multiagent reinforcement learning by payoff propagation. J Mach Learn Res 7:1789–1828

    MathSciNet  MATH  Google Scholar 

  19. 19.

    Konidaris G, Barto A (2009) Efficient skill learning using abstraction selection. In: International joint conference on artificial intelligence (IJCAI ’09), pp 1107–1112

  20. 20.

    Kravaris T, Spatharis C, Bastas A, Vouros GA, Blekas K, Andrienko G, Andrienko N, Garcia JM (2019) Resolving congestions in the air traffic management domain via multiagent reinforcement learning methods. Technical Report. arXiv preprint arXiv:1912.06860

  21. 21.

    Kravaris T, Vouros G, Spatharis C, Blekas K, Chalkiadakis G (2017) Learning policies for resolving demand–capacity imbalances during pre-tactical air traffic management. In: Multiagent system technologies—15th German conference (MATES‘17), pp 238–255

  22. 22.

    Kulkarni T, Narasimhan K, Saeedi A, Tenenbaum J (2016) Hierarchical deep reinforcement learning: Integrating temporal abstraction and intrinsic motivation. In: Advances in Neural Information Processing Systems (NIPS’16), pp 3675–3683

  23. 23.

    Li L, Walsh T, Littman M (2006) Towards a unified theory of state abstraction for MDPs. In: International symposium on artificial intelligence and mathematics (ISAIM‘06)

  24. 24.

    Ma A, Ouimet M, Cortés J (2020) Hierarchical reinforcement learning via dynamic subspace search for multi-agent planning. Auton Robot 44:485–503. https://doi.org/10.1007/s10514-019-09871-2

    Article  Google Scholar 

  25. 25.

    Makar R, Mahadevan S, Ghavamzadeh M (2001) Hierarchical multi-agent reinforcement learning. In: Proceedings of the fifth international conference on autonomous agents, AGENTS’01, pp 246–253

  26. 26.

    Malialis K, Delvin S, Kudenko D (2016) Resource abstraction for reinforcement learning in multiagent congestion problems. In: International conference on autonomous agents and multi-agent systems (AAMAS’16), pp 503–511

  27. 27.

    Mannor S, Menanche I, Hoze A, Klein U (2004) Dynamic abstraction in reinforcement learning via clustering. In: International conference on machine learning (ICML‘04). https://doi.org/10.1145/1015330.1015355

  28. 28.

    McGovern A, Barto A (2001) Automatic discovery of subgoals in reinforcement learning using diverse density. In: Proceedings of the eighteenth international conference on machine learning (ICML’01), pp 361–368

  29. 29.

    Meyers C (2006) Network flow problems and congestion games: complexity and approximation results. Ph.D. thesis, MIT

  30. 30.

    Milchtaich I (2004) Social optimality and cooperation in nonatomic congestion games. J Econ Theory 114(1):56–87

    MathSciNet  Article  Google Scholar 

  31. 31.

    Nachum T, Gu SS, Lee H, Levine S (2018) Data-efficient hierarchical reinforcement learning. In: 32nd Conference on neural information processing systems (NeurIPS 2018), pp 3303–3313

  32. 32.

    Parr R, Russell S (1998) Reinforcement learning with hierarchies of machines. In: Proceedings of the 1997 conference on advances in neural information processing systems (NIPS'97), vol 10, pp 1043–1049

  33. 33.

    Peng XB, Berseth G, Yin K, van de Panne M (2017) Deeploco: dynamic locomotion skills using hierarchical deep reinforcement learning. ACM Trans Graph 36(4):1–13. https://doi.org/10.1145/3072959.3073602

    Article  Google Scholar 

  34. 34.

    Penn M, Polukarov M, Tennenholtz M (2011) Congestion games with failures. Discr Appl Math 159(15):1508–1525

    MathSciNet  Article  Google Scholar 

  35. 35.

    Radulescu R, Vrancx P, Nowe A (2017) Analysing congestion problems in multi-agent reinforcement learning. In: Proceedings of the 16th conference on autonomous agents and multiagent systems (AAMAS’17), pp 1705–1707

  36. 36.

    Rasmussen D, Voelker A, Eliasmith C (2017) A neural model of hierarchical reinforcement learning. PLoS One 12(7):e0180234. https://doi.org/10.1371/journal.pone.0180234

    Article  Google Scholar 

  37. 37.

    Riemer M, Liu M, Tesauro G (2018) Learning abstract options. In: Proceedings of the 32nd international conference on neural information processing systems, NIPS’18, pp 10445-10455. Curran Associates Inc., Red Hook

  38. 38.

    Rosenthal RW (1973) A class of games processing pure-strategy nash equilibria. Int J Game Theory 2:65–67

    Article  Google Scholar 

  39. 39.

    Spatharis C, Blekas K, Bastas A, Kravaris T, Vouros GA (2019) Collaborative multiagent reinforcement learning schemes for air traffic management. In: 10th international conference on information, intelligence, systems and applications (IISA), pp 1–8

  40. 40.

    Spatharis C, Kravaris T, Vouros GA, Blekas K, Chalkadiakis G, Garcia JMC, Fernández EC (2018) Multiagent reinforcement learning methods to resolve demand capacity balance problems. In: Hellenic A.I. conference (SETN 2018), pp 2:1–2:9

  41. 41.

    Spatharis C, Kravaris T, Vouros GA, Blekas K, Cordero JMG (2018) Multiagent reinforcement learning methods for resolving demand—capacity imbalances. In: Digital avionics systems conference (DASC’18)

  42. 42.

    Sutton R, Precup D, Singh S (1999) Between MDPs and semi-MDPs: a framework for temporal abstraction in reinforcement learning. Artif Intell 112(1–2):181–211

    MathSciNet  Article  Google Scholar 

  43. 43.

    Tessler C, Givony S, Zahavy T, Mankowitz DJ, Mannor S (2017) A deep hierarchical approach to lifelong learning in minecraft. In: Proceedings of the thirty-first AAAI conference on artificial intelligence (AAAI '17), pp 1553–1561

  44. 44.

    Tumer K, Welch Z, Agogino A (2008) Aligning social welfare and agent preferences to alleviate traffic congestion. In: Proceedings of the 7th international joint conference on autonomous agents and multiagent systems (AAMAS’08), vol 2, pp 655–662

Download references


This work has been partially supported by the National Matching Funds 2017-2018 of the Greek Government and more specifically by the General Secretariat for Research and Technology (GSRT), related to DART(www.dart-research.eu) project. The major part of this work has been completed during the DART project where authors participated as members of the University of Piraeus Research Center group. We would like also to acknowledge the contribution of CRIDA for providing the data during the DART project.

Author information



Corresponding author

Correspondence to George A. Vouros.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Additional information

Publisher's Note

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

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Spatharis, C., Bastas, A., Kravaris, T. et al. Hierarchical multiagent reinforcement learning schemes for air traffic management. Neural Comput & Applic (2021). https://doi.org/10.1007/s00521-021-05748-7

Download citation


  • Multiagent reinforcement learning
  • Hierarchical learning
  • State abstraction
  • Congestion problems
  • Air traffic management