Imitating others by composition of primitive actions: A neuro-dynamic model

Hiroaki Arie, Takafumi Arakaki, Shigeki Sugano, Jun Tani

    Research output: Contribution to journalArticle

    19 Citations (Scopus)

    Abstract

    This paper introduces a novel neuro-dynamical model that accounts for possible mechanisms of action imitation and learning. It is considered that imitation learning requires at least two classes of generalization. One is generalization over sensorymotor trajectory variances, and the other class is on cognitive level which concerns on more qualitative understanding of compositional actions by own and others which do not necessarily depend on exact trajectories. This paper describes a possible model dealing with these classes of generalization by focusing on the problem of action compositionality. The model was evaluated in the experiments using a small humanoid robot. The robot was trained with a set of different actions concerning object manipulations which can be decomposed into sequences of action primitives. Then the robot was asked to imitate a novel compositional action demonstrated by a human subject which are composed from prior-learned action primitives. The results showed that the novel action can be successfully imitated by decomposing and composing it with the primitives by means of organizing unified intentional representation hosted by mirror neurons even though the trajectory-level appearance is different between the ones of observed and those of self-generated.

    Original languageEnglish
    Pages (from-to)729-741
    Number of pages13
    JournalRobotics and Autonomous Systems
    Volume60
    Issue number5
    DOIs
    Publication statusPublished - 2012 May

    Fingerprint

    Dynamic models
    Dynamic Model
    Trajectories
    Robots
    Chemical analysis
    Neurons
    Imitation
    Mirrors
    Trajectory
    Robot
    Compositionality
    Humanoid Robot
    Dynamical Model
    Experiments
    Manipulation
    Neuron
    Mirror
    Model
    Experiment
    Generalization

    Keywords

    • Cognitive robotics
    • Dynamical system
    • Imitation
    • Neural network

    ASJC Scopus subject areas

    • Control and Systems Engineering
    • Computer Science Applications
    • Software
    • Mathematics(all)

    Cite this

    Imitating others by composition of primitive actions : A neuro-dynamic model. / Arie, Hiroaki; Arakaki, Takafumi; Sugano, Shigeki; Tani, Jun.

    In: Robotics and Autonomous Systems, Vol. 60, No. 5, 05.2012, p. 729-741.

    Research output: Contribution to journalArticle

    Arie, Hiroaki ; Arakaki, Takafumi ; Sugano, Shigeki ; Tani, Jun. / Imitating others by composition of primitive actions : A neuro-dynamic model. In: Robotics and Autonomous Systems. 2012 ; Vol. 60, No. 5. pp. 729-741.
    @article{bd678a9f0b8e427da95069a05177b6b6,
    title = "Imitating others by composition of primitive actions: A neuro-dynamic model",
    abstract = "This paper introduces a novel neuro-dynamical model that accounts for possible mechanisms of action imitation and learning. It is considered that imitation learning requires at least two classes of generalization. One is generalization over sensorymotor trajectory variances, and the other class is on cognitive level which concerns on more qualitative understanding of compositional actions by own and others which do not necessarily depend on exact trajectories. This paper describes a possible model dealing with these classes of generalization by focusing on the problem of action compositionality. The model was evaluated in the experiments using a small humanoid robot. The robot was trained with a set of different actions concerning object manipulations which can be decomposed into sequences of action primitives. Then the robot was asked to imitate a novel compositional action demonstrated by a human subject which are composed from prior-learned action primitives. The results showed that the novel action can be successfully imitated by decomposing and composing it with the primitives by means of organizing unified intentional representation hosted by mirror neurons even though the trajectory-level appearance is different between the ones of observed and those of self-generated.",
    keywords = "Cognitive robotics, Dynamical system, Imitation, Neural network",
    author = "Hiroaki Arie and Takafumi Arakaki and Shigeki Sugano and Jun Tani",
    year = "2012",
    month = "5",
    doi = "10.1016/j.robot.2011.11.005",
    language = "English",
    volume = "60",
    pages = "729--741",
    journal = "Robotics and Autonomous Systems",
    issn = "0921-8890",
    publisher = "Elsevier",
    number = "5",

    }

    TY - JOUR

    T1 - Imitating others by composition of primitive actions

    T2 - A neuro-dynamic model

    AU - Arie, Hiroaki

    AU - Arakaki, Takafumi

    AU - Sugano, Shigeki

    AU - Tani, Jun

    PY - 2012/5

    Y1 - 2012/5

    N2 - This paper introduces a novel neuro-dynamical model that accounts for possible mechanisms of action imitation and learning. It is considered that imitation learning requires at least two classes of generalization. One is generalization over sensorymotor trajectory variances, and the other class is on cognitive level which concerns on more qualitative understanding of compositional actions by own and others which do not necessarily depend on exact trajectories. This paper describes a possible model dealing with these classes of generalization by focusing on the problem of action compositionality. The model was evaluated in the experiments using a small humanoid robot. The robot was trained with a set of different actions concerning object manipulations which can be decomposed into sequences of action primitives. Then the robot was asked to imitate a novel compositional action demonstrated by a human subject which are composed from prior-learned action primitives. The results showed that the novel action can be successfully imitated by decomposing and composing it with the primitives by means of organizing unified intentional representation hosted by mirror neurons even though the trajectory-level appearance is different between the ones of observed and those of self-generated.

    AB - This paper introduces a novel neuro-dynamical model that accounts for possible mechanisms of action imitation and learning. It is considered that imitation learning requires at least two classes of generalization. One is generalization over sensorymotor trajectory variances, and the other class is on cognitive level which concerns on more qualitative understanding of compositional actions by own and others which do not necessarily depend on exact trajectories. This paper describes a possible model dealing with these classes of generalization by focusing on the problem of action compositionality. The model was evaluated in the experiments using a small humanoid robot. The robot was trained with a set of different actions concerning object manipulations which can be decomposed into sequences of action primitives. Then the robot was asked to imitate a novel compositional action demonstrated by a human subject which are composed from prior-learned action primitives. The results showed that the novel action can be successfully imitated by decomposing and composing it with the primitives by means of organizing unified intentional representation hosted by mirror neurons even though the trajectory-level appearance is different between the ones of observed and those of self-generated.

    KW - Cognitive robotics

    KW - Dynamical system

    KW - Imitation

    KW - Neural network

    UR - http://www.scopus.com/inward/record.url?scp=84858450869&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84858450869&partnerID=8YFLogxK

    U2 - 10.1016/j.robot.2011.11.005

    DO - 10.1016/j.robot.2011.11.005

    M3 - Article

    AN - SCOPUS:84858450869

    VL - 60

    SP - 729

    EP - 741

    JO - Robotics and Autonomous Systems

    JF - Robotics and Autonomous Systems

    SN - 0921-8890

    IS - 5

    ER -