Efficient motor babbling using variance predictions from a recurrent neural network

Kuniyuki Takahashi, Kanata Suzuki, Tetsuya Ogata, Hadi Tjandra, Shigeki Sugano

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    1 Citation (Scopus)

    Abstract

    We propose an exploratory form of motor babbling that uses variance predictions from a recurrent neural network as a method to acquire the body dynamics of a robot with flexible joints. In conventional research methods, it is difficult to construct real robots because of the large number of motor babbling motions required. In motor babbling, different motions may be easy or difficult to predict. The variance is large in difficult-to-predict motions, whereas the variance is small in easy-topredict motions. We use a Stochastic Continuous Timescale Recurrent Neural Network to predict the accuracy and variance of motions. Using the proposed method, a robot can explore motions based on variance. To evaluate the proposed method, experiments were conducted in which the robot learns crank turning and door opening/closing tasks after exploring its body dynamics. The results show that the proposed method is capable of efficient motion generation for any given motion tasks.

    Original languageEnglish
    Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    PublisherSpringer Verlag
    Pages26-33
    Number of pages8
    Volume9491
    ISBN (Print)9783319265544
    DOIs
    Publication statusPublished - 2015
    Event22nd International Conference on Neural Information Processing, ICONIP 2015 - Istanbul, Turkey
    Duration: 2015 Nov 92015 Nov 12

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume9491
    ISSN (Print)03029743
    ISSN (Electronic)16113349

    Other

    Other22nd International Conference on Neural Information Processing, ICONIP 2015
    CountryTurkey
    CityIstanbul
    Period15/11/915/11/12

    Fingerprint

    Prediction Variance
    Recurrent neural networks
    Recurrent Neural Networks
    Robots
    Motion
    Robot
    Plant shutdowns
    Predict
    Research Methods
    Time Scales
    Experiments

    Keywords

    • Flexible joint robot
    • Motor babbling
    • Recurrent neural network

    ASJC Scopus subject areas

    • Computer Science(all)
    • Theoretical Computer Science

    Cite this

    Takahashi, K., Suzuki, K., Ogata, T., Tjandra, H., & Sugano, S. (2015). Efficient motor babbling using variance predictions from a recurrent neural network. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9491, pp. 26-33). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9491). Springer Verlag. https://doi.org/10.1007/978-3-319-26555-1_4

    Efficient motor babbling using variance predictions from a recurrent neural network. / Takahashi, Kuniyuki; Suzuki, Kanata; Ogata, Tetsuya; Tjandra, Hadi; Sugano, Shigeki.

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9491 Springer Verlag, 2015. p. 26-33 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9491).

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Takahashi, K, Suzuki, K, Ogata, T, Tjandra, H & Sugano, S 2015, Efficient motor babbling using variance predictions from a recurrent neural network. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 9491, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9491, Springer Verlag, pp. 26-33, 22nd International Conference on Neural Information Processing, ICONIP 2015, Istanbul, Turkey, 15/11/9. https://doi.org/10.1007/978-3-319-26555-1_4
    Takahashi K, Suzuki K, Ogata T, Tjandra H, Sugano S. Efficient motor babbling using variance predictions from a recurrent neural network. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9491. Springer Verlag. 2015. p. 26-33. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-26555-1_4
    Takahashi, Kuniyuki ; Suzuki, Kanata ; Ogata, Tetsuya ; Tjandra, Hadi ; Sugano, Shigeki. / Efficient motor babbling using variance predictions from a recurrent neural network. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9491 Springer Verlag, 2015. pp. 26-33 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
    @inproceedings{a467980bcbac4c4da62543be9ed0eb93,
    title = "Efficient motor babbling using variance predictions from a recurrent neural network",
    abstract = "We propose an exploratory form of motor babbling that uses variance predictions from a recurrent neural network as a method to acquire the body dynamics of a robot with flexible joints. In conventional research methods, it is difficult to construct real robots because of the large number of motor babbling motions required. In motor babbling, different motions may be easy or difficult to predict. The variance is large in difficult-to-predict motions, whereas the variance is small in easy-topredict motions. We use a Stochastic Continuous Timescale Recurrent Neural Network to predict the accuracy and variance of motions. Using the proposed method, a robot can explore motions based on variance. To evaluate the proposed method, experiments were conducted in which the robot learns crank turning and door opening/closing tasks after exploring its body dynamics. The results show that the proposed method is capable of efficient motion generation for any given motion tasks.",
    keywords = "Flexible joint robot, Motor babbling, Recurrent neural network",
    author = "Kuniyuki Takahashi and Kanata Suzuki and Tetsuya Ogata and Hadi Tjandra and Shigeki Sugano",
    year = "2015",
    doi = "10.1007/978-3-319-26555-1_4",
    language = "English",
    isbn = "9783319265544",
    volume = "9491",
    series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
    publisher = "Springer Verlag",
    pages = "26--33",
    booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

    }

    TY - GEN

    T1 - Efficient motor babbling using variance predictions from a recurrent neural network

    AU - Takahashi, Kuniyuki

    AU - Suzuki, Kanata

    AU - Ogata, Tetsuya

    AU - Tjandra, Hadi

    AU - Sugano, Shigeki

    PY - 2015

    Y1 - 2015

    N2 - We propose an exploratory form of motor babbling that uses variance predictions from a recurrent neural network as a method to acquire the body dynamics of a robot with flexible joints. In conventional research methods, it is difficult to construct real robots because of the large number of motor babbling motions required. In motor babbling, different motions may be easy or difficult to predict. The variance is large in difficult-to-predict motions, whereas the variance is small in easy-topredict motions. We use a Stochastic Continuous Timescale Recurrent Neural Network to predict the accuracy and variance of motions. Using the proposed method, a robot can explore motions based on variance. To evaluate the proposed method, experiments were conducted in which the robot learns crank turning and door opening/closing tasks after exploring its body dynamics. The results show that the proposed method is capable of efficient motion generation for any given motion tasks.

    AB - We propose an exploratory form of motor babbling that uses variance predictions from a recurrent neural network as a method to acquire the body dynamics of a robot with flexible joints. In conventional research methods, it is difficult to construct real robots because of the large number of motor babbling motions required. In motor babbling, different motions may be easy or difficult to predict. The variance is large in difficult-to-predict motions, whereas the variance is small in easy-topredict motions. We use a Stochastic Continuous Timescale Recurrent Neural Network to predict the accuracy and variance of motions. Using the proposed method, a robot can explore motions based on variance. To evaluate the proposed method, experiments were conducted in which the robot learns crank turning and door opening/closing tasks after exploring its body dynamics. The results show that the proposed method is capable of efficient motion generation for any given motion tasks.

    KW - Flexible joint robot

    KW - Motor babbling

    KW - Recurrent neural network

    UR - http://www.scopus.com/inward/record.url?scp=84952053276&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84952053276&partnerID=8YFLogxK

    U2 - 10.1007/978-3-319-26555-1_4

    DO - 10.1007/978-3-319-26555-1_4

    M3 - Conference contribution

    SN - 9783319265544

    VL - 9491

    T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

    SP - 26

    EP - 33

    BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

    PB - Springer Verlag

    ER -