A hypothetical neural network model for generation of human precision grip

Yuki Moritani, Naomichi Ogihara

Research output: Contribution to journalArticle

Abstract

Humans can stably hold and skillfully manipulate an object by coordinated control of a complex, redundant musculoskeletal system. However, how the human central nervous system actually accomplishes precision grip tasks by coordinated control of fingertip forces remains unclear. In the present study, we aimed to construct a hypothetical neural network model that can spontaneously generate humanlike precision grip. The nervous system was modeled as a recurrent neural network model prescribing kinematic and kinetic constraints that must be satisfied in precision grip tasks in the form of energy functions. The recurrent neural network autonomously behaves so as to decrease the energy functions; therefore, given the estimated mass and center-of-mass location of the target object, the nervous system model can spontaneously generate muscle activation signals that achieve stable precision grips due to dynamic relaxation of the energy functions embedded in the nervous system. Fingertip forces are modulated by sensory information about slip between the object and fingertips. A two-dimensional musculoskeletal model of the human hand with a thumb and an index finger was constructed. Forward dynamic simulation of the precision grip was performed using the proposed neural network model. Our results demonstrated that the proposed neural network model could stably pinch and successfully hold up the object in various conditions, including changes in friction, object shape, object mass, and center-of-mass location. The proposed hypothetical neuro-computational model may possibly explain some aspects of the control strategy humans use for precision grip.

Original languageEnglish
Pages (from-to)213-224
Number of pages12
JournalNeural Networks
Volume110
DOIs
Publication statusPublished - 2019 Feb 1

Fingerprint

Neural Networks (Computer)
Hand Strength
Neural networks
Neurology
Nervous System
Recurrent neural networks
Musculoskeletal System
Musculoskeletal system
Friction
Thumb
Biomechanical Phenomena
Fingers
Central Nervous System
Hand
Muscle
Kinematics
Chemical activation
Muscles
Kinetics
Computer simulation

Keywords

  • Grasping
  • Hand
  • Motor control
  • Optimization
  • Simulation

ASJC Scopus subject areas

  • Cognitive Neuroscience
  • Artificial Intelligence

Cite this

A hypothetical neural network model for generation of human precision grip. / Moritani, Yuki; Ogihara, Naomichi.

In: Neural Networks, Vol. 110, 01.02.2019, p. 213-224.

Research output: Contribution to journalArticle

Moritani, Yuki ; Ogihara, Naomichi. / A hypothetical neural network model for generation of human precision grip. In: Neural Networks. 2019 ; Vol. 110. pp. 213-224.
@article{fc25657c7a9d4e4f9831c0c1f924db3e,
title = "A hypothetical neural network model for generation of human precision grip",
abstract = "Humans can stably hold and skillfully manipulate an object by coordinated control of a complex, redundant musculoskeletal system. However, how the human central nervous system actually accomplishes precision grip tasks by coordinated control of fingertip forces remains unclear. In the present study, we aimed to construct a hypothetical neural network model that can spontaneously generate humanlike precision grip. The nervous system was modeled as a recurrent neural network model prescribing kinematic and kinetic constraints that must be satisfied in precision grip tasks in the form of energy functions. The recurrent neural network autonomously behaves so as to decrease the energy functions; therefore, given the estimated mass and center-of-mass location of the target object, the nervous system model can spontaneously generate muscle activation signals that achieve stable precision grips due to dynamic relaxation of the energy functions embedded in the nervous system. Fingertip forces are modulated by sensory information about slip between the object and fingertips. A two-dimensional musculoskeletal model of the human hand with a thumb and an index finger was constructed. Forward dynamic simulation of the precision grip was performed using the proposed neural network model. Our results demonstrated that the proposed neural network model could stably pinch and successfully hold up the object in various conditions, including changes in friction, object shape, object mass, and center-of-mass location. The proposed hypothetical neuro-computational model may possibly explain some aspects of the control strategy humans use for precision grip.",
keywords = "Grasping, Hand, Motor control, Optimization, Simulation",
author = "Yuki Moritani and Naomichi Ogihara",
year = "2019",
month = "2",
day = "1",
doi = "10.1016/j.neunet.2018.12.001",
language = "English",
volume = "110",
pages = "213--224",
journal = "Neural Networks",
issn = "0893-6080",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - A hypothetical neural network model for generation of human precision grip

AU - Moritani, Yuki

AU - Ogihara, Naomichi

PY - 2019/2/1

Y1 - 2019/2/1

N2 - Humans can stably hold and skillfully manipulate an object by coordinated control of a complex, redundant musculoskeletal system. However, how the human central nervous system actually accomplishes precision grip tasks by coordinated control of fingertip forces remains unclear. In the present study, we aimed to construct a hypothetical neural network model that can spontaneously generate humanlike precision grip. The nervous system was modeled as a recurrent neural network model prescribing kinematic and kinetic constraints that must be satisfied in precision grip tasks in the form of energy functions. The recurrent neural network autonomously behaves so as to decrease the energy functions; therefore, given the estimated mass and center-of-mass location of the target object, the nervous system model can spontaneously generate muscle activation signals that achieve stable precision grips due to dynamic relaxation of the energy functions embedded in the nervous system. Fingertip forces are modulated by sensory information about slip between the object and fingertips. A two-dimensional musculoskeletal model of the human hand with a thumb and an index finger was constructed. Forward dynamic simulation of the precision grip was performed using the proposed neural network model. Our results demonstrated that the proposed neural network model could stably pinch and successfully hold up the object in various conditions, including changes in friction, object shape, object mass, and center-of-mass location. The proposed hypothetical neuro-computational model may possibly explain some aspects of the control strategy humans use for precision grip.

AB - Humans can stably hold and skillfully manipulate an object by coordinated control of a complex, redundant musculoskeletal system. However, how the human central nervous system actually accomplishes precision grip tasks by coordinated control of fingertip forces remains unclear. In the present study, we aimed to construct a hypothetical neural network model that can spontaneously generate humanlike precision grip. The nervous system was modeled as a recurrent neural network model prescribing kinematic and kinetic constraints that must be satisfied in precision grip tasks in the form of energy functions. The recurrent neural network autonomously behaves so as to decrease the energy functions; therefore, given the estimated mass and center-of-mass location of the target object, the nervous system model can spontaneously generate muscle activation signals that achieve stable precision grips due to dynamic relaxation of the energy functions embedded in the nervous system. Fingertip forces are modulated by sensory information about slip between the object and fingertips. A two-dimensional musculoskeletal model of the human hand with a thumb and an index finger was constructed. Forward dynamic simulation of the precision grip was performed using the proposed neural network model. Our results demonstrated that the proposed neural network model could stably pinch and successfully hold up the object in various conditions, including changes in friction, object shape, object mass, and center-of-mass location. The proposed hypothetical neuro-computational model may possibly explain some aspects of the control strategy humans use for precision grip.

KW - Grasping

KW - Hand

KW - Motor control

KW - Optimization

KW - Simulation

UR - http://www.scopus.com/inward/record.url?scp=85059121494&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85059121494&partnerID=8YFLogxK

U2 - 10.1016/j.neunet.2018.12.001

DO - 10.1016/j.neunet.2018.12.001

M3 - Article

C2 - 30597446

AN - SCOPUS:85059121494

VL - 110

SP - 213

EP - 224

JO - Neural Networks

JF - Neural Networks

SN - 0893-6080

ER -