Traffic3d: A rich 3D-traffic environment to train intelligent agents

Deepeka Garg*, Maria Chli, George Vogiatzis

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The last few years marked a substantial development in the domain of Deep Reinforcement Learning. However, a crucial and not yet fully achieved objective is to devise intelligent agents which can be successfully taken out of the laboratory and employed in the real world. Intelligent agents that are successfully deployable in true physical settings, require substantial prior exposure to their intended environments. When this is not practical or possible, the agents benefit from being trained and tested on powerful test-beds, effectively replicating the real world. To achieve traffic management at an unprecedented level of efficiency, in this paper, we introduce a significantly richer new traffic simulation environment; Traffic3D. Traffic3D is a unique platform built to effectively simulate and evaluate a variety of 3D-road traffic scenarios, closely mimicking real-world traffic characteristics including faithful simulation of individual vehicle behavior, precise physics of movement and photo-realism. We discuss the merits of Traffic3D in comparison to state-of-the-art traffic-based simulators. Along with deep reinforcement learning, Traffic3D facilitates research across various domains such as object detection and segmentation, unsupervised representation learning, visual question answering, procedural generation, imitation learning and learning by interaction.

Original languageEnglish
Title of host publicationComputational Science - ICCS 2019 - 19th International Conference, 2019, Proceedings
EditorsJoão M.F. Rodrigues, Pedro J.S. Cardoso, Jânio Monteiro, Roberto Lam, Valeria V. Krzhizhanovskaya, Michael H. Lees, Peter M.A. Sloot, Jack J. Dongarra
PublisherSpringer-Verlag Wien
Pages749-755
Number of pages7
ISBN (Print)9783030227494
DOIs
Publication statusPublished - 8 Jun 2019
Event19th International Conference on Computational Science, ICCS 2019 - Faro, Portugal
Duration: 12 Jun 201914 Jun 2019

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11540 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference19th International Conference on Computational Science, ICCS 2019
CountryPortugal
CityFaro
Period12/06/1914/06/19

Fingerprint

Intelligent agents
Intelligent Agents
Reinforcement learning
Traffic
Reinforcement Learning
Traffic Simulation
Traffic Management
Imitation
Physics
Question Answering
Simulators
Object Detection
Simulation Environment
Faithful
Testbed
Simulator
Segmentation
Scenarios
Evaluate
Interaction

Bibliographical note

© Springer Nature B.V. 2019. The final publication is available at Springer via http://dx.doi.org/10.1007/978-3-030-22750-0_74

Keywords

  • Deep learning
  • Intelligent transportation systems
  • Machine learning
  • Virtual reality 3D traffic simulator

Cite this

Garg, D., Chli, M., & Vogiatzis, G. (2019). Traffic3d: A rich 3D-traffic environment to train intelligent agents. In J. M. F. Rodrigues, P. J. S. Cardoso, J. Monteiro, R. Lam, V. V. Krzhizhanovskaya, M. H. Lees, P. M. A. Sloot, ... J. J. Dongarra (Eds.), Computational Science - ICCS 2019 - 19th International Conference, 2019, Proceedings (pp. 749-755). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11540 LNCS). Springer-Verlag Wien. https://doi.org/10.1007/978-3-030-22750-0_74
Garg, Deepeka ; Chli, Maria ; Vogiatzis, George. / Traffic3d : A rich 3D-traffic environment to train intelligent agents. Computational Science - ICCS 2019 - 19th International Conference, 2019, Proceedings. editor / João M.F. Rodrigues ; Pedro J.S. Cardoso ; Jânio Monteiro ; Roberto Lam ; Valeria V. Krzhizhanovskaya ; Michael H. Lees ; Peter M.A. Sloot ; Jack J. Dongarra. Springer-Verlag Wien, 2019. pp. 749-755 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{f58ee15384f846778ac137f6db438c73,
title = "Traffic3d: A rich 3D-traffic environment to train intelligent agents",
abstract = "The last few years marked a substantial development in the domain of Deep Reinforcement Learning. However, a crucial and not yet fully achieved objective is to devise intelligent agents which can be successfully taken out of the laboratory and employed in the real world. Intelligent agents that are successfully deployable in true physical settings, require substantial prior exposure to their intended environments. When this is not practical or possible, the agents benefit from being trained and tested on powerful test-beds, effectively replicating the real world. To achieve traffic management at an unprecedented level of efficiency, in this paper, we introduce a significantly richer new traffic simulation environment; Traffic3D. Traffic3D is a unique platform built to effectively simulate and evaluate a variety of 3D-road traffic scenarios, closely mimicking real-world traffic characteristics including faithful simulation of individual vehicle behavior, precise physics of movement and photo-realism. We discuss the merits of Traffic3D in comparison to state-of-the-art traffic-based simulators. Along with deep reinforcement learning, Traffic3D facilitates research across various domains such as object detection and segmentation, unsupervised representation learning, visual question answering, procedural generation, imitation learning and learning by interaction.",
keywords = "Deep learning, Intelligent transportation systems, Machine learning, Virtual reality 3D traffic simulator",
author = "Deepeka Garg and Maria Chli and George Vogiatzis",
note = "{\circledC} Springer Nature B.V. 2019. The final publication is available at Springer via http://dx.doi.org/10.1007/978-3-030-22750-0_74",
year = "2019",
month = "6",
day = "8",
doi = "10.1007/978-3-030-22750-0_74",
language = "English",
isbn = "9783030227494",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer-Verlag Wien",
pages = "749--755",
editor = "Rodrigues, {Jo{\~a}o M.F.} and Cardoso, {Pedro J.S.} and J{\^a}nio Monteiro and Roberto Lam and Krzhizhanovskaya, {Valeria V.} and Lees, {Michael H.} and Sloot, {Peter M.A.} and Dongarra, {Jack J.}",
booktitle = "Computational Science - ICCS 2019 - 19th International Conference, 2019, Proceedings",
address = "Austria",

}

Garg, D, Chli, M & Vogiatzis, G 2019, Traffic3d: A rich 3D-traffic environment to train intelligent agents. in JMF Rodrigues, PJS Cardoso, J Monteiro, R Lam, VV Krzhizhanovskaya, MH Lees, PMA Sloot & JJ Dongarra (eds), Computational Science - ICCS 2019 - 19th International Conference, 2019, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11540 LNCS, Springer-Verlag Wien, pp. 749-755, 19th International Conference on Computational Science, ICCS 2019, Faro, Portugal, 12/06/19. https://doi.org/10.1007/978-3-030-22750-0_74

Traffic3d : A rich 3D-traffic environment to train intelligent agents. / Garg, Deepeka; Chli, Maria; Vogiatzis, George.

Computational Science - ICCS 2019 - 19th International Conference, 2019, Proceedings. ed. / João M.F. Rodrigues; Pedro J.S. Cardoso; Jânio Monteiro; Roberto Lam; Valeria V. Krzhizhanovskaya; Michael H. Lees; Peter M.A. Sloot; Jack J. Dongarra. Springer-Verlag Wien, 2019. p. 749-755 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11540 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Traffic3d

T2 - A rich 3D-traffic environment to train intelligent agents

AU - Garg, Deepeka

AU - Chli, Maria

AU - Vogiatzis, George

N1 - © Springer Nature B.V. 2019. The final publication is available at Springer via http://dx.doi.org/10.1007/978-3-030-22750-0_74

PY - 2019/6/8

Y1 - 2019/6/8

N2 - The last few years marked a substantial development in the domain of Deep Reinforcement Learning. However, a crucial and not yet fully achieved objective is to devise intelligent agents which can be successfully taken out of the laboratory and employed in the real world. Intelligent agents that are successfully deployable in true physical settings, require substantial prior exposure to their intended environments. When this is not practical or possible, the agents benefit from being trained and tested on powerful test-beds, effectively replicating the real world. To achieve traffic management at an unprecedented level of efficiency, in this paper, we introduce a significantly richer new traffic simulation environment; Traffic3D. Traffic3D is a unique platform built to effectively simulate and evaluate a variety of 3D-road traffic scenarios, closely mimicking real-world traffic characteristics including faithful simulation of individual vehicle behavior, precise physics of movement and photo-realism. We discuss the merits of Traffic3D in comparison to state-of-the-art traffic-based simulators. Along with deep reinforcement learning, Traffic3D facilitates research across various domains such as object detection and segmentation, unsupervised representation learning, visual question answering, procedural generation, imitation learning and learning by interaction.

AB - The last few years marked a substantial development in the domain of Deep Reinforcement Learning. However, a crucial and not yet fully achieved objective is to devise intelligent agents which can be successfully taken out of the laboratory and employed in the real world. Intelligent agents that are successfully deployable in true physical settings, require substantial prior exposure to their intended environments. When this is not practical or possible, the agents benefit from being trained and tested on powerful test-beds, effectively replicating the real world. To achieve traffic management at an unprecedented level of efficiency, in this paper, we introduce a significantly richer new traffic simulation environment; Traffic3D. Traffic3D is a unique platform built to effectively simulate and evaluate a variety of 3D-road traffic scenarios, closely mimicking real-world traffic characteristics including faithful simulation of individual vehicle behavior, precise physics of movement and photo-realism. We discuss the merits of Traffic3D in comparison to state-of-the-art traffic-based simulators. Along with deep reinforcement learning, Traffic3D facilitates research across various domains such as object detection and segmentation, unsupervised representation learning, visual question answering, procedural generation, imitation learning and learning by interaction.

KW - Deep learning

KW - Intelligent transportation systems

KW - Machine learning

KW - Virtual reality 3D traffic simulator

UR - http://www.scopus.com/inward/record.url?scp=85068448575&partnerID=8YFLogxK

UR - https://link.springer.com/chapter/10.1007%2F978-3-030-22750-0_74

U2 - 10.1007/978-3-030-22750-0_74

DO - 10.1007/978-3-030-22750-0_74

M3 - Conference contribution

AN - SCOPUS:85068448575

SN - 9783030227494

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 749

EP - 755

BT - Computational Science - ICCS 2019 - 19th International Conference, 2019, Proceedings

A2 - Rodrigues, João M.F.

A2 - Cardoso, Pedro J.S.

A2 - Monteiro, Jânio

A2 - Lam, Roberto

A2 - Krzhizhanovskaya, Valeria V.

A2 - Lees, Michael H.

A2 - Sloot, Peter M.A.

A2 - Dongarra, Jack J.

PB - Springer-Verlag Wien

ER -

Garg D, Chli M, Vogiatzis G. Traffic3d: A rich 3D-traffic environment to train intelligent agents. In Rodrigues JMF, Cardoso PJS, Monteiro J, Lam R, Krzhizhanovskaya VV, Lees MH, Sloot PMA, Dongarra JJ, editors, Computational Science - ICCS 2019 - 19th International Conference, 2019, Proceedings. Springer-Verlag Wien. 2019. p. 749-755. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-22750-0_74