講演抄録/キーワード |
講演名 |
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 |
抄録 |
(和) |
(まだ登録されていません) |
(英) |
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. |
キーワード |
(和) |
/ / / / / / / |
(英) |
Rapid Transit System Simulation / Reinforcement Learning based Supervisory Control / / / / / / |
文献情報 |
信学技報, vol. 117, no. 431, ITS2017-66, pp. 29-33, 2018年2月. |
資料番号 |
ITS2017-66 |
発行日 |
2018-02-08 (ITS, IE) |
ISSN |
Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 |
著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します.(許諾番号:10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
PDFダウンロード |
ITS2017-66 IE2017-98 |
|