Online Deep Reinforcement Learning on Assigned Weight Spaghetti Grasping in One Time using Soft Actor-Critic

DOI
  • Gamolped Prem
    Department of Mechanical Information Science and Technology, Kyushu Institute of Technology
  • Chumkamon Sakmongkon
    Department of Mechanical Information Science and Technology, Kyushu Institute of Technology
  • Piyavichyanon Chanapol
    Department of Creative Informatics, Kyushu Institute of Technology
  • Hayashi Eiji
    Department of Mechanical Information Science and Technology, Kyushu Institute of Technology
  • Mowshowitz Abbe
    Department of Computer Science, The City College of New York

説明

Artificial Intelligence and Robotics have become essential and widely used to package food. Packaging an assigned weight spaghetti into a lunch box at one time can be difficult. This paper proposes a solution for one-time grasping using Deep Reinforcement Learning (DRL) based on the Soft Actor-Critic algorithm on the manipulator. Spaghetti detection and segmentation are implemented from the RGB-D camera for the observation. We conclude that the experiment shows the effectively grasped result can almost succeed within 10% of the target weight in the experimental environment.

収録刊行物

キーワード

詳細情報 詳細情報について

  • CRID
    1390010292579077504
  • DOI
    10.5954/icarob.2022.os19-1
  • ISSN
    21887829
  • 本文言語コード
    en
  • データソース種別
    • JaLC
    • Crossref
    • OpenAIRE
  • 抄録ライセンスフラグ
    使用不可

問題の指摘

ページトップへ