Abstract
In the area of autonomous driving there is a need to flexibly configure and simulate more complex individual pedestrian behavior in critical traffic scenes which goes beyond predefined behavior simulation. This paper presents a novel human-oriented, agent-based pedestrian simulation framework, named HAIL, that addresses this challenge. HAIL allows to simulate human pedestrian behavior through means of imitation learning by virtual agents. For this purpose, HAIL combines the 3D traffic simulation environment OpenDS with an integrated imitation learning environment and hybrid agents with AJAN. For predictive behavior planning on the tactical and strategical level, AJAN is extended with Answer Set Programming. For pedestrian behavior imitation learning on the operational level, HAIL utilizes the module InfoSalGAIL for generation of pedestrian paths learned from demonstration by its human counterpart as expert. Among others, an application example has been demonstrated that HAIL can be applied to solve a common challenge in the Neural Network domain, namely the out-of-distribution (OOD), e.g. never shown scenarios would raise an uncertainty prediction level, by unison work of the two different behavior generation frameworks.
Keywords
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsNotes
- 1.
Carla: https://carla.org/.
- 2.
LGSVL: https://www.lgsvlsimulator.com/.
- 3.
OpenDS - open source driving simulation: https://opends.dfki.de/.
- 4.
- 5.
VADERE Crowd simulation: http://www.vadere.org/.
- 6.
PDESIM - pedestrian crowd simulation: http://pedsim.silmaril.org/.
- 7.
- 8.
AJAN-service: https://github.com/aantakli/AJAN-service AJAN-editor: https://github.com/aantakli/AJAN-editor.
- 9.
Not like EKs, where only the corresponding agent behavior has access to.
- 10.
AJAN uses LibGDX-BTs: https://github.com/libgdx/gdx-ai/wiki/Behavior-Trees.
- 11.
Upon initialization, an AJAN agent receives RDF descriptions of available InfoSalGAIL models defining trained street configurations.
- 12.
- 13.
In AJAN we use the Potassco clingo solver: https://potassco.org/clingo/.
- 14.
Videos of the FoV of the pedestrian agent in scenario A can be found at: https://cloud.dfki.de/owncloud/index.php/s/HAf5wQtMAx3F9K5.
References
Abbeel, P., Ng, A.Y.: Apprenticeship learning via inverse reinforcement learning. In: Proceedings 21st International Conference Machine Learning (ICML) (2004)
Antakli, A., et al.: Agent-based web supported simulation of human-robot collaboration. In: Proceedings International Conference Web Information Systems and Technologies (WebIST) (2019)
Antakli, A., Zinnikus, I., Klusch, M.: ASP-driven BDI-planning agents in virtual 3D environments. In: Klusch, M., Unland, R., Shehory, O., Pokahr, A., Ahrndt, S. (eds.) MATES 2016. LNCS (LNAI), vol. 9872, pp. 198–214. Springer, Cham (2016). https://doi.org/10.1007/978-3-319-45889-2_15
Bloem, M., Bambos, N.: Infinite time horizon maximum causal entropy inverse reinforcement learning. In: International Conference Decision and Control. IEEE (2014)
Brambilla, M., Facca, F.M.: Building semantic web portals with a model-driven design approach. In: Web Technologies: Concepts, Methodologies, Tools, and Applications, pp. 541–570. IGI Global (2010)
Chen, X., Duan, Y., Houthooft, R., Schulman, J., Sutskever, I., Abbeel, P.: InfoGAN: interpretable representation learning by information maximizing generative adversarial nets. In: Proceedings 29th International Conference Neural Information Processing Systems (NeurIPS), pp. 2172–2180 (2016)
Colledanchise, M., Ögren, P.: Behavior Trees in Robotics and AI: An Introduction. CRC Press, Boca Raton (2018)
Ho, J., Ermon, S.: Generative adversarial imitation learning. In: Proceedings 29th International Conference Neural Information Processing Systems (NeurIPS) (2016)
Ianni, G., Martello, A., Panetta, C., Terracina, G.: Efficiently querying RDF (S) ontologies with answer set programming. Logic Comput. 19(4), 671–695 (2009)
Ishaque, M.M., Noland, R.B.: Behavioural issues in pedestrian speed choice and street crossing behaviour: a review. Transp. Rev. 28(1), 61–85 (2008)
Karamouzas, I., Heil, P., van Beek, P., Overmars, M.H.: A predictive collision avoidance model for pedestrian simulation. In: Egges, A., Geraerts, R., Overmars, M. (eds.) MIG 2009. LNCS, vol. 5884, pp. 41–52. Springer, Heidelberg (2009). https://doi.org/10.1007/978-3-642-10347-6_4
Lee, J., Li, T., De Vos, M., Padget, J.: Using social institutions to guide virtual agent behaviour. In: The AAMAS Workshop on Cognitive Agents for Virtual Environments (CAVE-2013). Citeseer (2013)
Li, Y., Song, J., Ermon, S.: InfoGAIL: interpretable imitation learning from visual demonstrations. In: Proceedings 30th International Conference Neural Information Processing Systems (NeurIPS) (2017)
Panagiotidi, S., Nieves, J.C., Vazquez-Salceda, J.: A framework to model norm dynamics in answer set programming. In: Proceedings International Workshop Multi-agent Logics, Languages, and Organisations (MALLOW), vol. CEUR 494 (2009)
Paxton, C., Hundt, A., Jonathan, F., Guerin, K., Hager, G.D.: CoSTAR: instructing collaborative robots with behavior trees and vision. In: Proceedings IEEE International Conference Robotics and Automation (ICRA). IEEE (2017)
Reynolds, C.W.: Steering behaviors for autonomous characters. In: Game Developers Conference, vol. 1999, pp. 763–782. Citeseer (1999)
Ross, S., Bagnell, D.: Efficient reductions for imitation learning. In: Proceedings 13th International Conference Artificial Intelligence and Statistics (2010)
Teknomo, K., Takeyama, Y., Inamura, H.: Review on microscopic pedestrian simulation model. arXiv preprint arXiv:1609.01808 (2016)
Vizzari, G., Crociani, L., Bandini, S.: An agent-based model for plausible wayfinding in pedestrian simulation. Eng. Appl. Artif. Intell. 87, 103241 (2020)
Vozniak, I., Klusch, M., Antakli, A., Müller, C.: InfoSalGAIL: visual attention-empowered imitation learning of pedestrians in critical TRAFC scenarios. In: International Conference Neural Computation Theory and Application (NCTA). IEEE (2020)
Acknowledgements
The work described in this paper has been funded by the German Federal Ministry of Education and Research (BMBF) through the project REACT (grant no. 01/W17003) and in part by Huawei Munich Research Center.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this paper
Cite this paper
Antakli, A., Vozniak, I., Lipp, N., Klusch, M., Müller, C. (2021). HAIL: Modular Agent-Based Pedestrian Imitation Learning. In: Dignum, F., Corchado, J.M., De La Prieta, F. (eds) Advances in Practical Applications of Agents, Multi-Agent Systems, and Social Good. The PAAMS Collection. PAAMS 2021. Lecture Notes in Computer Science(), vol 12946. Springer, Cham. https://doi.org/10.1007/978-3-030-85739-4_3
Download citation
DOI: https://doi.org/10.1007/978-3-030-85739-4_3
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-85738-7
Online ISBN: 978-3-030-85739-4
eBook Packages: Computer ScienceComputer Science (R0)