• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Automatic crystal growth by visuomotor learning

Research Project

Project/Area Number 21H01681
Research Category

Grant-in-Aid for Scientific Research (B)

Allocation TypeSingle-year Grants
Section一般
Review Section Basic Section 26060:Metals production and resources production-related
Research InstitutionNagoya University

Principal Investigator

Harada Shunta  名古屋大学, 未来材料・システム研究所, 准教授 (30612460)

Project Period (FY) 2021-04-01 – 2024-03-31
Project Status Completed (Fiscal Year 2023)
Budget Amount *help
¥18,070,000 (Direct Cost: ¥13,900,000、Indirect Cost: ¥4,170,000)
Fiscal Year 2023: ¥2,340,000 (Direct Cost: ¥1,800,000、Indirect Cost: ¥540,000)
Fiscal Year 2022: ¥2,470,000 (Direct Cost: ¥1,900,000、Indirect Cost: ¥570,000)
Fiscal Year 2021: ¥13,260,000 (Direct Cost: ¥10,200,000、Indirect Cost: ¥3,060,000)
Keywords単結晶育成 / 浮遊帯域溶融法(FZ法) / 強化学習 / ガウス混合モデル(GMM) / プロセスインフォマティクス / 自動化 / 結晶成長 / 浮遊帯域溶融法 / 混合ガウスモデル / 製造自動化 / 逆強化学習 / 機械学習 / 自動操業 / 深層視覚運動学習
Outline of Research at the Start

ロボット制御で用いられつつある深層視覚運動学習を、浮遊帯域溶融(FZ)法による単結晶育成の制御に応用し、「職人技」を定量化し、熟練のオペレータの制御を学習することにより、「見て」「判断し」「制御する」アルゴリズムを構築する。これにより熟練のオペレータの操業を模倣し、「職人技」を超える制御を実現する。

Outline of Final Research Achievements

In this study, we aimed to automate single crystal growth using the Floating Zone (FZ) method by developing a control system based on reinforcement learning. Specifically, we estimated the state of the molten zone from observation images and built algorithms based on the operational data of skilled operators. We constructed a state transition model using the Gaussian Mixture Model (GMM) and developed a control model using the Proximal Policy Optimization (PPO) algorithm. As a result, we achieved improved quality and high-precision automatic control of FZ crystal growth. Further demonstration experiments with actual equipment are planned for the future.

Academic Significance and Societal Importance of the Research Achievements

学術的意義:本研究は、強化学習を用いて浮遊帯域溶融法(FZ法)の単結晶育成を自動化する新たなアプローチを提供し、製造プロセスの効率化と高精度化に寄与します。ガウス混合モデル(GMM)とProximal Policy Optimization(PPO)アルゴリズムの融合により、結晶成長の理論的理解と実践的応用の架け橋となります。
社会的意義:本研究の成果により、高品質な単結晶の安定供給が可能となり、半導体デバイスの性能向上や製造コストの削減が期待されます。また、製造現場における熟練技術者の不足問題を緩和し、製造業全体の競争力強化に貢献します。

Report

(4 results)
  • 2023 Annual Research Report   Final Research Report ( PDF )
  • 2022 Annual Research Report
  • 2021 Annual Research Report
  • Research Products

    (10 results)

All 2023 2022 2021 Other

All Journal Article (3 results) (of which Peer Reviewed: 3 results,  Open Access: 3 results) Presentation (6 results) (of which Int'l Joint Research: 2 results,  Invited: 2 results) Remarks (1 results)

  • [Journal Article] Data-driven automated control algorithm for floating-zone crystal growth derived by reinforcement learning2023

    • Author(s)
      Tosa Yusuke、Omae Ryo、Matsumoto Ryohei、Sumitani Shogo、Harada Shunta
    • Journal Title

      Scientific Reports

      Volume: 13 Issue: 1 Pages: 7517-7517

    • DOI

      10.1038/s41598-023-34732-5

    • Related Report
      2023 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Current Status and Prospects of AI Applications for Quality Control2022

    • Author(s)
      炭谷 翔悟, 西田 陽良, 原田 俊太
    • Journal Title

      SYSTEMS, CONTROL AND INFORMATION

      Volume: 66 Issue: 5 Pages: 161-166

    • DOI

      10.11509/isciesci.66.5_161

    • ISSN
      0916-1600, 2424-1806
    • Year and Date
      2022-05-15
    • Related Report
      2022 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Prediction of operating dynamics in floating-zone crystal growth using Gaussian mixture model2022

    • Author(s)
      R. Omae, S. Sumitani, Y. Tosa, S. Harada
    • Journal Title

      Science and Technology of Advanced Materials: Methods

      Volume: 2 Issue: 1 Pages: 294-301

    • DOI

      10.1080/27660400.2022.2107884

    • Related Report
      2022 Annual Research Report
    • Peer Reviewed / Open Access
  • [Presentation] 製造プロセスの制御への強化学習の応用ー浮遊帯域溶融法による結晶成長を例に2023

    • Author(s)
      原田俊太
    • Organizer
      半導体の結晶成長と加工および評価に関する産学連携委員会
    • Related Report
      2023 Annual Research Report
    • Invited
  • [Presentation] Data-Driven Automated Control Algorithm for Floating-Zone Crystal Growth Using Reinforcement Learning2023

    • Author(s)
      Shunta Harada, Yusuke Tosa, Ryo Omae, Ryohei Matsumoto, Shogo Sumitani
    • Organizer
      ICCGE-20
    • Related Report
      2023 Annual Research Report
    • Int'l Joint Research / Invited
  • [Presentation] Data-Driven Modeling and Adaptive Optimization of Floating Zone Crystal Growth Process Applying Gaussian Mixture Model and Reinforcement Learning2023

    • Author(s)
      Shunta Harada, Yusuke Tosa, Ryo Omae, Ryohei Matsumoto, Shogo Sumitani
    • Organizer
      2023 MRS Spring Meeting
    • Related Report
      2023 Annual Research Report
    • Int'l Joint Research
  • [Presentation] 混合ガウスモデルと強化学習による浮遊帯域溶融法による結晶育成自動化の検討2022

    • Author(s)
      原田俊太, 土佐祐介, 大前遼, 松本遼平, 炭谷翔悟
    • Organizer
      第51回結晶成長国内会議
    • Related Report
      2022 Annual Research Report
  • [Presentation] 強化学習を用いた浮遊帯域溶融法による結晶成長の自動制御モデルの構築2022

    • Author(s)
      土佐 祐介, 炭谷 翔悟, 大前 遼, 原田 俊太
    • Organizer
      第69回応用物理学会春季学術講演会
    • Related Report
      2021 Annual Research Report
  • [Presentation] 混合ガウスモデルを用いた浮遊帯域溶融法における融液状態のダイナミクス推定2021

    • Author(s)
      大前 遼, 炭谷 翔悟, 土佐 祐介, 原田 俊太
    • Organizer
      第82回応用物理学会秋季学術講演会
    • Related Report
      2021 Annual Research Report
  • [Remarks] Google scholar Shunta Harada

    • URL

      https://scholar.google.co.jp/citations?user=gr7I_cUAAAAJ&hl=ja

    • Related Report
      2023 Annual Research Report

URL: 

Published: 2021-04-28   Modified: 2025-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi