Route optimization of construction machine by deep reinforcement learning

Shunya Tanabe, Zeyuan Sun, Masayuki Nakatani, Yutaka Uchimura

Research output: Contribution to journalArticle

Abstract

After it was reported that an AI player scored higher in Atari2600 games than skilled human players by using deep reinforcement learning techniques, many researchers were inspired to apply deep reinforcement leaning in various fields. This paper focuses on the autonomous ground leveling work by a bulldozer, which is expected to optimize the action of the bulldozer. In a previous work, we implemented a deep Q learning method by giving the images as the input data for the network. However, when learning the image using the convolution layer as the input using deep reinforcement learning, it requires a large computational cost for the learning process. If the size of the neural network is shrunken by contriving the data to be supplied to the input, the learning time (duration) will be reduced. This paper describes the comparison results for different orders of input data. the transition of the learning sequence is also evaluated.

Original languageEnglish
Pages (from-to)401-408
Number of pages8
JournalIEEJ Transactions on Industry Applications
Volume139
Issue number4
DOIs
Publication statusPublished - 2019 Jan 1

Fingerprint

Reinforcement learning
Convolution
Reinforcement
Neural networks
Costs

Keywords

  • Artificial intelligence
  • Autonomous control
  • Deep reinforcement learning
  • Leveling machine
  • Machine learning

ASJC Scopus subject areas

  • Industrial and Manufacturing Engineering
  • Electrical and Electronic Engineering

Cite this

Route optimization of construction machine by deep reinforcement learning. / Tanabe, Shunya; Sun, Zeyuan; Nakatani, Masayuki; Uchimura, Yutaka.

In: IEEJ Transactions on Industry Applications, Vol. 139, No. 4, 01.01.2019, p. 401-408.

Research output: Contribution to journalArticle

Tanabe, Shunya ; Sun, Zeyuan ; Nakatani, Masayuki ; Uchimura, Yutaka. / Route optimization of construction machine by deep reinforcement learning. In: IEEJ Transactions on Industry Applications. 2019 ; Vol. 139, No. 4. pp. 401-408.
@article{69f32b24d4de466ebbcf3326d3bce3f5,
title = "Route optimization of construction machine by deep reinforcement learning",
abstract = "After it was reported that an AI player scored higher in Atari2600 games than skilled human players by using deep reinforcement learning techniques, many researchers were inspired to apply deep reinforcement leaning in various fields. This paper focuses on the autonomous ground leveling work by a bulldozer, which is expected to optimize the action of the bulldozer. In a previous work, we implemented a deep Q learning method by giving the images as the input data for the network. However, when learning the image using the convolution layer as the input using deep reinforcement learning, it requires a large computational cost for the learning process. If the size of the neural network is shrunken by contriving the data to be supplied to the input, the learning time (duration) will be reduced. This paper describes the comparison results for different orders of input data. the transition of the learning sequence is also evaluated.",
keywords = "Artificial intelligence, Autonomous control, Deep reinforcement learning, Leveling machine, Machine learning",
author = "Shunya Tanabe and Zeyuan Sun and Masayuki Nakatani and Yutaka Uchimura",
year = "2019",
month = "1",
day = "1",
doi = "10.1541/ieejias.139.401",
language = "English",
volume = "139",
pages = "401--408",
journal = "IEEJ Transactions on Industry Applications",
issn = "0913-6339",
publisher = "The Institute of Electrical Engineers of Japan",
number = "4",

}

TY - JOUR

T1 - Route optimization of construction machine by deep reinforcement learning

AU - Tanabe, Shunya

AU - Sun, Zeyuan

AU - Nakatani, Masayuki

AU - Uchimura, Yutaka

PY - 2019/1/1

Y1 - 2019/1/1

N2 - After it was reported that an AI player scored higher in Atari2600 games than skilled human players by using deep reinforcement learning techniques, many researchers were inspired to apply deep reinforcement leaning in various fields. This paper focuses on the autonomous ground leveling work by a bulldozer, which is expected to optimize the action of the bulldozer. In a previous work, we implemented a deep Q learning method by giving the images as the input data for the network. However, when learning the image using the convolution layer as the input using deep reinforcement learning, it requires a large computational cost for the learning process. If the size of the neural network is shrunken by contriving the data to be supplied to the input, the learning time (duration) will be reduced. This paper describes the comparison results for different orders of input data. the transition of the learning sequence is also evaluated.

AB - After it was reported that an AI player scored higher in Atari2600 games than skilled human players by using deep reinforcement learning techniques, many researchers were inspired to apply deep reinforcement leaning in various fields. This paper focuses on the autonomous ground leveling work by a bulldozer, which is expected to optimize the action of the bulldozer. In a previous work, we implemented a deep Q learning method by giving the images as the input data for the network. However, when learning the image using the convolution layer as the input using deep reinforcement learning, it requires a large computational cost for the learning process. If the size of the neural network is shrunken by contriving the data to be supplied to the input, the learning time (duration) will be reduced. This paper describes the comparison results for different orders of input data. the transition of the learning sequence is also evaluated.

KW - Artificial intelligence

KW - Autonomous control

KW - Deep reinforcement learning

KW - Leveling machine

KW - Machine learning

UR - http://www.scopus.com/inward/record.url?scp=85063755922&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85063755922&partnerID=8YFLogxK

U2 - 10.1541/ieejias.139.401

DO - 10.1541/ieejias.139.401

M3 - Article

AN - SCOPUS:85063755922

VL - 139

SP - 401

EP - 408

JO - IEEJ Transactions on Industry Applications

JF - IEEJ Transactions on Industry Applications

SN - 0913-6339

IS - 4

ER -