Use of Action Label in Deep Predictive Learning for Robot Manipulation

被引:0
|
作者
Kase, Kei [1 ,2 ]
Utsumi, Chikara [1 ,2 ]
Domae, Yukiyasu [2 ]
Ogata, Tetsuya [1 ,2 ]
机构
[1] Waseda Univ, Tokyo, Japan
[2] Natl Inst Adv Ind Sci & Technol, Tokyo, Japan
来源
2022 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS) | 2022年
关键词
D O I
10.1109/IROS47612.2022.9982091
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Various forms of human knowledge can be explicitly used to enhance deep robot learning from demonstrations. Annotation of subtasks from task segmentation is one type of human symbolism and knowledge. Annotated subtasks can be referred to as action labels, which are more primitive symbols that can be building blocks for more complex human reasoning, like language instructions. However, action labels are not widely used to boost learning processes because of problems that include (1) real-time annotation for online manipulation, (2) temporal inconsistency by annotators, (3) difference in data characteristics of motor commands and action labels, and (4) annotation cost. To address these problems, we propose the Gated Action Motor Predictive Learning (GAMPL) framework to leverage action labels for improved performance. GAMPL has two modules to obtain soft action labels compatible with motor commands and to generate motion. In this study, GAMPL is evaluated for towel-folding manipulation tasks in a real environment with a six degrees-of-freedom (6 DoF) robot and shows improved generalizability with action labels.
引用
收藏
页码:13459 / 13465
页数:7
相关论文
共 50 条
  • [41] Model Predictive Actor-Critic: Accelerating Robot Skill Acquisition with Deep Reinforcement Learning
    Morgan, Andrew S.
    Nandha, Daljeet
    Chalvatzaki, Georgia
    D'Eramo, Carlo
    Dollar, Aaron M.
    Peters, Jan
    2021 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION (ICRA 2021), 2021, : 6672 - 6678
  • [42] Deep label embedding learning for classification
    Nousi, Paraskevi
    Tefas, Anastasios
    APPLIED SOFT COMPUTING, 2024, 163
  • [43] Deep Learning with Label Differential Privacy
    Ghazi, Badih
    Golowich, Noah
    Kumar, Ravi
    Manurangsi, Pasin
    Zhang, Chiyuan
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 34 (NEURIPS 2021), 2021, 34
  • [44] A Comparison of Action Spaces for Learning Manipulation Tasks
    Varin, Patrick
    Grossman, Lev
    Kuindersma, Scott
    2019 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS), 2019, : 6015 - 6021
  • [45] MULTI-LABEL DEEP ACTIVE LEARNING WITH LABEL CORRELATION
    Ranganathan, Hiranmayi
    Venkateswara, Hemanth
    Chakraborty, Shayok
    Panchanathan, Sethuraman
    2018 25TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP), 2018, : 3418 - 3422
  • [46] Learning Action-oriented Grasping for Manipulation
    Ud Din, Muhayy
    Sarwar, M. Usman
    Zahoor, Imran
    Qazi, Wajahat M.
    Rosell, Jan
    2019 24TH IEEE INTERNATIONAL CONFERENCE ON EMERGING TECHNOLOGIES AND FACTORY AUTOMATION (ETFA), 2019, : 1575 - 1578
  • [47] Learning Whole-Body Manipulation for Quadrupedal Robot
    Jeon, Seunghun
    Jung, Moonkyu
    Choi, Suyoung
    Kim, Beomjoon
    Hwangbo, Jemin
    IEEE ROBOTICS AND AUTOMATION LETTERS, 2024, 9 (01) : 699 - 706
  • [48] Learning Robust Policies for Object Manipulation with Robot Swarms
    Gebhardt, Gregor H. W.
    Daun, Kevin
    Schnaubelt, Marius
    Neumann, Gerhard
    2018 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION (ICRA), 2018, : 7688 - 7695
  • [49] A Review of Robot Learning for Manipulation: Challenges, Representations, and Algorithms
    Kroemer, Oliver
    Niekum, Scott
    Konidaris, George
    JOURNAL OF MACHINE LEARNING RESEARCH, 2021, 22
  • [50] Learning the Elasticity Parameters of Deformable Objects with a Manipulation Robot
    Frank, Barbara
    Schmedding, Ruediger
    Stachniss, Cyrill
    Teschner, Matthias
    Burgard, Wolfram
    IEEE/RSJ 2010 INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS 2010), 2010, : 1877 - 1883