IKEMOTO Shuhei

写真a

Title

Associate Professor

Laboratory

2-4 Hibikino, Wakamatsu-ku, Kitakyushu-shi, Fukuoka

Research Fields, Keywords

Robotics

Undergraduate Education 【 display / non-display

  • 2005.03   Kanazawa University   Faculty of Engineering   Graduated   JAPAN

  • 2003.03   Toyota National College of Technology     Graduated   JAPAN

Post Graduate Education 【 display / non-display

  • 2010.03  Osaka University  Graduate School, Division of Engineering  Doctoral Program  Completed  JAPAN

Degree 【 display / non-display

  • Osaka University -  Doctor of Engineering  2010.03

Biography in Kyutech 【 display / non-display

  • 2019.04
    -
    Now

    Kyushu Institute of TechnologyGraduate School of Life Science and Systems Engineering   Department of Human Intelligence Systems   Associate Professor  

Biography before Kyutech 【 display / non-display

  • 2015.04
    -
    2019.03

      Assistant Professor   JAPAN

  • 2014.07
    -
    2015.03

      Specially Appointed Assistant Professor   JAPAN

  • 2010.04
    -
    2014.06

      Assistant Professor   JAPAN

 

Publications (Article) 【 display / non-display

  • Reconstructing state-space from movie using convolutional autoencoder for robot control

      867   480 - 489   2019.01  [Refereed]

     View Summary

    © 2019, Springer Nature Switzerland AG. In contrast with intensive studies for hardware development in soft robotics, approaches to construct a controller for soft robots has been relying on heuristics. One of the biggest reasons of this issue is that even reconstructing the state-space to describe the behavior is difficult. In this study, we propose a method to reconstruct state-space from movies using a convolutional autoencoder for robot control. In the proposed method, the process that reduces the number of dimensions of each frame in movies is regulated by additional losses making latent variables orthogonal each other and apt to model the forward dynamics. The proposed method was successfully validated through a simulation where a two links planar manipulator is modeled using the movie and controlled based on the forward model.

    DOI Scopus

  • Modular robot that modeled cell membrane dynamics of a cellular slime mold

      867   302 - 313   2019.01  [Refereed]

     View Summary

    © 2019, Springer Nature Switzerland AG. Understanding of the design principles for implementing adaptive functions with respect to engineering currently remains stalled in the conceptual level. However, living organisms exhibit great adaptive function by skillfully relating shape and function in a spatio-temporal manner. In this study, we focus on amoeboid organisms because these organisms have a variable morphology that relates shape and function. Amoeboid organisms in the natural world (i.e., cellular slime molds) locomote through changing the cell membrane shape by inducing the internal protoplasmic streaming. Based on this mechanism, we developed modular robots that modeled the cell membrane dynamics of a cellar slime mold.

    DOI Scopus

  • Highly-Integrated Muscle-Spindles for Pneumatic Artificial Muscles Made from Conductive Fabrics

      11556 LNAI   171 - 182   2019.01  [Refereed]

     View Summary

    © 2019, Springer Nature Switzerland AG. Pneumatic artificial muscles (PAMs) actuating bio-inspired structures are widely used to mimic the human musculoskeletal system. The research in this field can improve the understanding of the human’s physical abilities through a close recreation of its natural structure. This paper will introduce an enhancement to PAMs resembling the sensory-feedback of the muscles spindles’ group Ia and II afferent neurons. The artificial muscle spindle presented in this paper is embedded into the muscle and wraps around like its biological counterpart. Previous publications of artificial muscle spindles mostly aimed to output a signal, which correlates to the length of the muscle. This approach, however, aimed to recreate the natural muscle spindle as close as possible in regards to its functional principal and positioning inside of the PAM. By using conductive fabrics, a deeply embedded sensor type was created, which feedback correlates to the expansion of the PAM’s inner tube, as well as the pressure in-between of the inner tube and the outer braided sleeve. In this paper, a constructional approach on a biomimetic muscle spindle is introduced, including its Ia and II afferent neurons.

    DOI Scopus

  • Common dimensional autoencoder for identifying agonist-antagonist muscle pairs in musculoskeletal robots

      867   325 - 333   2019.01  [Refereed]

     View Summary

    © 2019, Springer Nature Switzerland AG. One of the distinctive features of musculoskeletal systems is the redundancy provided by agonist-antagonist muscle pairs. To identify agonist-antagonist muscle pairs in a musculoskeletal robot, however, is difficult as it requires complex structures to mimic human physiology. Thus, we propose a method to identify agonist-antagonist muscle pairs in a complex musculoskeletal robot using motor commands. Moreover, the common dimensional autoencoder, where the encoded feature has identical dimensions to the original input vector, is used to separate the image and the kernel spaces for each time period. Finally, we successfully confirmed the efficacy of our method by applying a 2-link planar manipulator to a 3-pairs-6-muscles configuration.

    DOI Scopus

  • Optimal Feedback Control Based on Analytical Linear Models Extracted from Neural Networks Trained for Nonlinear Systems

        8689 - 8694   2018.12  [Refereed]

     View Summary

    © 2018 IEEE. A number of researches have been focusing on the development and control of robots with soft structures such as flexible musculoskeletal systems. Thus far, it has been reported that these robots can achieve high adaptability to environments despite their extremely simple controllers. However, because these robots are difficult to model mathematically, there is still no systematic design policy, in which control theory has been playing a role in conventional robotics, for constituting simple controllers. To tackle this problem, we propose a new approach using a neural network to obtain mathematical models. In particular, with this method, the control theory is applied to linear system models extracted from a network trained to express the forward dynamics of a robot. Through simulations, the validity and advantage of the proposed method was successfully confirmed.

    DOI Scopus

display all >>