Paper Abstract and Keywords |
Presentation |
2018-02-15 11:30
Stochastic Discrete Event Simulation Environment for Autonomous Cart Fleet for Artificial Intelligent Training and Reinforcement Learning Algorithms Naohisa Hashimoto, Ali Boyali, Shin Kato (AIST), Takao Otsuka, Kazuhisa Mizushima, Manabu Omae (Keio Univ) ITS2017-66 IE2017-98 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
In this report we give details of a Discrete Event Simulation (DES) framework coded in Python environment for simulation and analysis of a customized Personal Rapid Transport (PRT) with passenger behavior. The prior analysis of the system is a must before deployment of the autonomous PRT cars (carts) to make decision of initial investment, number of carts required and to design supervisory control algorithms that reduces the defined cost functional in the optimization. The simulation program coded in consideration of training Artificial Intelligent (AI) agents by Deep Reinforcement Learning methods similar to OpenAI Gym environment. The basic requirements for modeling of discreet stochastic simulation and analyses are summarized in the report. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Rapid Transit System Simulation / Reinforcement Learning based Supervisory Control / / / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 117, no. 431, ITS2017-66, pp. 29-33, Feb. 2018. |
Paper # |
ITS2017-66 |
Date of Issue |
2018-02-08 (ITS, IE) |
ISSN |
Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 |
Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
Download PDF |
ITS2017-66 IE2017-98 |
|