Development of a neuromorphic control system for a lightweight humanoid robot

Michele Folgheraiter, Amina Keldibek, Bauyrzhan Aubakir, Shyngys Salakchinov, Giuseppina Gini, Alessio Mauro Franchi, Matteo Bana

Research output: Contribution to journalConference article

1 Citation (Scopus)

Abstract

A neuromorphic control system for a lightweight middle size humanoid biped robot built using 3D printing techniques is proposed. The control architecture consists of different modules capable to learn and autonomously reproduce complex periodic trajectories. Each module is represented by a chaotic Recurrent Neural Network (RNN) with a core of dynamic neurons randomly and sparsely connected with fixed synapses. A set of read-out units with adaptable synapses realize a linear combination of the neurons output in order to reproduce the target signals. Different experiments were conducted to find out the optimal initialization for the RNN's parameters. From simulation results, using normalized signals obtained from the robot model, it was proven that all the instances of the control module can learn and reproduce the target trajectories with an average RMS error of 1.63 and variance 0.74.

Original languageEnglish
Article number012021
JournalIOP Conference Series: Materials Science and Engineering
Volume185
Issue number1
DOIs
Publication statusPublished - Mar 22 2017
Event1st Annual International Conference on Information Technology and Digital Applications 2016, ICITDA 2016 - Yogyakarta, Indonesia
Duration: Nov 14 2016Nov 16 2016

Fingerprint

Neurons
Trajectories
Robots
Control systems
Recurrent neural networks
Printing
Experiments

ASJC Scopus subject areas

  • Materials Science(all)
  • Engineering(all)

Cite this

Folgheraiter, M., Keldibek, A., Aubakir, B., Salakchinov, S., Gini, G., Franchi, A. M., & Bana, M. (2017). Development of a neuromorphic control system for a lightweight humanoid robot. IOP Conference Series: Materials Science and Engineering, 185(1), [012021]. https://doi.org/10.1088/1757-899X/185/1/012021

Development of a neuromorphic control system for a lightweight humanoid robot. / Folgheraiter, Michele; Keldibek, Amina; Aubakir, Bauyrzhan; Salakchinov, Shyngys; Gini, Giuseppina; Franchi, Alessio Mauro; Bana, Matteo.

In: IOP Conference Series: Materials Science and Engineering, Vol. 185, No. 1, 012021, 22.03.2017.

Research output: Contribution to journalConference article

Folgheraiter, M, Keldibek, A, Aubakir, B, Salakchinov, S, Gini, G, Franchi, AM & Bana, M 2017, 'Development of a neuromorphic control system for a lightweight humanoid robot', IOP Conference Series: Materials Science and Engineering, vol. 185, no. 1, 012021. https://doi.org/10.1088/1757-899X/185/1/012021
Folgheraiter, Michele ; Keldibek, Amina ; Aubakir, Bauyrzhan ; Salakchinov, Shyngys ; Gini, Giuseppina ; Franchi, Alessio Mauro ; Bana, Matteo. / Development of a neuromorphic control system for a lightweight humanoid robot. In: IOP Conference Series: Materials Science and Engineering. 2017 ; Vol. 185, No. 1.
@article{5b15c6a6552b4d79bb8da6005ca63f5f,
title = "Development of a neuromorphic control system for a lightweight humanoid robot",
abstract = "A neuromorphic control system for a lightweight middle size humanoid biped robot built using 3D printing techniques is proposed. The control architecture consists of different modules capable to learn and autonomously reproduce complex periodic trajectories. Each module is represented by a chaotic Recurrent Neural Network (RNN) with a core of dynamic neurons randomly and sparsely connected with fixed synapses. A set of read-out units with adaptable synapses realize a linear combination of the neurons output in order to reproduce the target signals. Different experiments were conducted to find out the optimal initialization for the RNN's parameters. From simulation results, using normalized signals obtained from the robot model, it was proven that all the instances of the control module can learn and reproduce the target trajectories with an average RMS error of 1.63 and variance 0.74.",
author = "Michele Folgheraiter and Amina Keldibek and Bauyrzhan Aubakir and Shyngys Salakchinov and Giuseppina Gini and Franchi, {Alessio Mauro} and Matteo Bana",
year = "2017",
month = "3",
day = "22",
doi = "10.1088/1757-899X/185/1/012021",
language = "English",
volume = "185",
journal = "IOP Conference Series: Materials Science and Engineering",
issn = "1757-8981",
publisher = "IOP Publishing Ltd.",
number = "1",

}

TY - JOUR

T1 - Development of a neuromorphic control system for a lightweight humanoid robot

AU - Folgheraiter, Michele

AU - Keldibek, Amina

AU - Aubakir, Bauyrzhan

AU - Salakchinov, Shyngys

AU - Gini, Giuseppina

AU - Franchi, Alessio Mauro

AU - Bana, Matteo

PY - 2017/3/22

Y1 - 2017/3/22

N2 - A neuromorphic control system for a lightweight middle size humanoid biped robot built using 3D printing techniques is proposed. The control architecture consists of different modules capable to learn and autonomously reproduce complex periodic trajectories. Each module is represented by a chaotic Recurrent Neural Network (RNN) with a core of dynamic neurons randomly and sparsely connected with fixed synapses. A set of read-out units with adaptable synapses realize a linear combination of the neurons output in order to reproduce the target signals. Different experiments were conducted to find out the optimal initialization for the RNN's parameters. From simulation results, using normalized signals obtained from the robot model, it was proven that all the instances of the control module can learn and reproduce the target trajectories with an average RMS error of 1.63 and variance 0.74.

AB - A neuromorphic control system for a lightweight middle size humanoid biped robot built using 3D printing techniques is proposed. The control architecture consists of different modules capable to learn and autonomously reproduce complex periodic trajectories. Each module is represented by a chaotic Recurrent Neural Network (RNN) with a core of dynamic neurons randomly and sparsely connected with fixed synapses. A set of read-out units with adaptable synapses realize a linear combination of the neurons output in order to reproduce the target signals. Different experiments were conducted to find out the optimal initialization for the RNN's parameters. From simulation results, using normalized signals obtained from the robot model, it was proven that all the instances of the control module can learn and reproduce the target trajectories with an average RMS error of 1.63 and variance 0.74.

UR - http://www.scopus.com/inward/record.url?scp=85017475993&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85017475993&partnerID=8YFLogxK

U2 - 10.1088/1757-899X/185/1/012021

DO - 10.1088/1757-899X/185/1/012021

M3 - Conference article

VL - 185

JO - IOP Conference Series: Materials Science and Engineering

JF - IOP Conference Series: Materials Science and Engineering

SN - 1757-8981

IS - 1

M1 - 012021

ER -