Presentation | 2021-05-14 A Study of Reinforcement Learning for Solving Multi-Vehicle Routing Problems Kazuaki Akashi, Shunsuke Kanai, Kenichi Tayama, Zhao Wang, Yuusuke Nakano, Ken Nishimatsu, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | In recent years, deep reinforcement learning method have been studied for solving vehicle routing problems. For multi-vehicle routing problems, multi-agent methods have been proposed in which each vehicle is an agent, but these methods are limited in patterns of routes that can be generated because the order in which the vehicles select their destinations is fixed. Therefore, in this paper, we propose a single-agent method in which the agent is the operator who plans the destination of each vehicle. As a result of a simple simulation, the proposed method can remove the limitation of conventional methods, and can generate the route for more cases than the Google OR-tools. On the other hand, the average cost of the routes generated by the proposed method is inferior to the Google OR-tools, so we have to improve the performance by changing the model. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Vehicle Routing Problems / Reinforcement Learning / Neural network / Attention / AI |
Paper # | ICM2021-5 |
Date of Issue | 2021-05-06 (ICM) |
Conference Information | |
Committee | ICM / IPSJ-CSEC / IPSJ-IOT |
---|---|
Conference Date | 2021/5/13(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Online |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | Kazuhiko Kinoshita(Tokushima Univ.) |
Vice Chair | Haruo Ooishi(NTT) / Eiji Takahashi(NEC) |
Secretary | Haruo Ooishi(BOSCO Technologies) / Eiji Takahashi(Fujitsu Lab.) |
Assistant | Yoshifumi Kato(NTT) |
Paper Information | |
Registration To | Technical Committee on Information and Communication Management / Special Interest Group on Computer Security / Special Interest Group on Internet and Operation Technology |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | A Study of Reinforcement Learning for Solving Multi-Vehicle Routing Problems |
Sub Title (in English) | |
Keyword(1) | Vehicle Routing Problems |
Keyword(2) | Reinforcement Learning |
Keyword(3) | Neural network |
Keyword(4) | Attention |
Keyword(5) | AI |
1st Author's Name | Kazuaki Akashi |
1st Author's Affiliation | NTT Access Network Service Systems Laboratories(NTT) |
2nd Author's Name | Shunsuke Kanai |
2nd Author's Affiliation | NTT Access Network Service Systems Laboratories(NTT) |
3rd Author's Name | Kenichi Tayama |
3rd Author's Affiliation | NTT Access Network Service Systems Laboratories(NTT) |
4th Author's Name | Zhao Wang |
4th Author's Affiliation | NTT Network Technology Laboratories(NTT) |
5th Author's Name | Yuusuke Nakano |
5th Author's Affiliation | NTT Network Technology Laboratories(NTT) |
6th Author's Name | Ken Nishimatsu |
6th Author's Affiliation | NTT Network Technology Laboratories(NTT) |
Date | 2021-05-14 |
Paper # | ICM2021-5 |
Volume (vol) | vol.121 |
Number (no) | ICM-13 |
Page | pp.pp.23-28(ICM), |
#Pages | 6 |
Date of Issue | 2021-05-06 (ICM) |