Statistical dynamical systems for skills acquisition in humanoids

Sylvain Calinon, Zhibin Li, Tohid Alizadeh, Nikos G. Tsagarakis, Darwin G. Caldwell

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

89 Citations (Scopus)

Abstract

Learning by imitation in humanoids is challenging due to the unpredictable environments these robots have to face during reproduction. Two sets of tools are relevant for this purpose: 1) probabilistic machine learning methods that can extract and exploit the regularities and important features of the task; and 2) dynamical systems that can cope with perturbation in real-time without having to replan the whole movement. We present a learning by imitation approach combining the two benefits. It is based on a superposition of virtual spring-damper systems to drive a humanoid robot's movement. The method relies on a statistical description of the springs attractor points acting in different candidate frames of reference. It extends dynamic movement primitives models by formulating the dynamical systems parameters estimation problem as a Gaussian mixture regression problem with projection in different coordinate systems. The robot exploits local variability information extracted from multiple demonstrations of movements to determine which frames are relevant for the task, and how the movement should be modulated with respect to these frames. The approach is tested on the new prototype of the COMAN compliant humanoid with time-based and time-invariant movements, including bimanual coordination skills.

Original languageEnglish
Title of host publicationIEEE-RAS International Conference on Humanoid Robots
Pages323-329
Number of pages7
DOIs
Publication statusPublished - 2012
Externally publishedYes
Event2012 12th IEEE-RAS International Conference on Humanoid Robots, Humanoids 2012 - Osaka, Japan
Duration: Nov 29 2012Dec 1 2012

Other

Other2012 12th IEEE-RAS International Conference on Humanoid Robots, Humanoids 2012
CountryJapan
CityOsaka
Period11/29/1212/1/12

Fingerprint

Dynamical systems
Robots
Parameter estimation
Learning systems
Demonstrations

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Vision and Pattern Recognition
  • Hardware and Architecture
  • Human-Computer Interaction
  • Electrical and Electronic Engineering

Cite this

Calinon, S., Li, Z., Alizadeh, T., Tsagarakis, N. G., & Caldwell, D. G. (2012). Statistical dynamical systems for skills acquisition in humanoids. In IEEE-RAS International Conference on Humanoid Robots (pp. 323-329). [6651539] https://doi.org/10.1109/HUMANOIDS.2012.6651539

Statistical dynamical systems for skills acquisition in humanoids. / Calinon, Sylvain; Li, Zhibin; Alizadeh, Tohid; Tsagarakis, Nikos G.; Caldwell, Darwin G.

IEEE-RAS International Conference on Humanoid Robots. 2012. p. 323-329 6651539.

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

Calinon, S, Li, Z, Alizadeh, T, Tsagarakis, NG & Caldwell, DG 2012, Statistical dynamical systems for skills acquisition in humanoids. in IEEE-RAS International Conference on Humanoid Robots., 6651539, pp. 323-329, 2012 12th IEEE-RAS International Conference on Humanoid Robots, Humanoids 2012, Osaka, Japan, 11/29/12. https://doi.org/10.1109/HUMANOIDS.2012.6651539
Calinon S, Li Z, Alizadeh T, Tsagarakis NG, Caldwell DG. Statistical dynamical systems for skills acquisition in humanoids. In IEEE-RAS International Conference on Humanoid Robots. 2012. p. 323-329. 6651539 https://doi.org/10.1109/HUMANOIDS.2012.6651539
Calinon, Sylvain ; Li, Zhibin ; Alizadeh, Tohid ; Tsagarakis, Nikos G. ; Caldwell, Darwin G. / Statistical dynamical systems for skills acquisition in humanoids. IEEE-RAS International Conference on Humanoid Robots. 2012. pp. 323-329
@inproceedings{136b0f7af2b34f30a6a051596517e934,
title = "Statistical dynamical systems for skills acquisition in humanoids",
abstract = "Learning by imitation in humanoids is challenging due to the unpredictable environments these robots have to face during reproduction. Two sets of tools are relevant for this purpose: 1) probabilistic machine learning methods that can extract and exploit the regularities and important features of the task; and 2) dynamical systems that can cope with perturbation in real-time without having to replan the whole movement. We present a learning by imitation approach combining the two benefits. It is based on a superposition of virtual spring-damper systems to drive a humanoid robot's movement. The method relies on a statistical description of the springs attractor points acting in different candidate frames of reference. It extends dynamic movement primitives models by formulating the dynamical systems parameters estimation problem as a Gaussian mixture regression problem with projection in different coordinate systems. The robot exploits local variability information extracted from multiple demonstrations of movements to determine which frames are relevant for the task, and how the movement should be modulated with respect to these frames. The approach is tested on the new prototype of the COMAN compliant humanoid with time-based and time-invariant movements, including bimanual coordination skills.",
author = "Sylvain Calinon and Zhibin Li and Tohid Alizadeh and Tsagarakis, {Nikos G.} and Caldwell, {Darwin G.}",
year = "2012",
doi = "10.1109/HUMANOIDS.2012.6651539",
language = "English",
isbn = "9781467313698",
pages = "323--329",
booktitle = "IEEE-RAS International Conference on Humanoid Robots",

}

TY - GEN

T1 - Statistical dynamical systems for skills acquisition in humanoids

AU - Calinon, Sylvain

AU - Li, Zhibin

AU - Alizadeh, Tohid

AU - Tsagarakis, Nikos G.

AU - Caldwell, Darwin G.

PY - 2012

Y1 - 2012

N2 - Learning by imitation in humanoids is challenging due to the unpredictable environments these robots have to face during reproduction. Two sets of tools are relevant for this purpose: 1) probabilistic machine learning methods that can extract and exploit the regularities and important features of the task; and 2) dynamical systems that can cope with perturbation in real-time without having to replan the whole movement. We present a learning by imitation approach combining the two benefits. It is based on a superposition of virtual spring-damper systems to drive a humanoid robot's movement. The method relies on a statistical description of the springs attractor points acting in different candidate frames of reference. It extends dynamic movement primitives models by formulating the dynamical systems parameters estimation problem as a Gaussian mixture regression problem with projection in different coordinate systems. The robot exploits local variability information extracted from multiple demonstrations of movements to determine which frames are relevant for the task, and how the movement should be modulated with respect to these frames. The approach is tested on the new prototype of the COMAN compliant humanoid with time-based and time-invariant movements, including bimanual coordination skills.

AB - Learning by imitation in humanoids is challenging due to the unpredictable environments these robots have to face during reproduction. Two sets of tools are relevant for this purpose: 1) probabilistic machine learning methods that can extract and exploit the regularities and important features of the task; and 2) dynamical systems that can cope with perturbation in real-time without having to replan the whole movement. We present a learning by imitation approach combining the two benefits. It is based on a superposition of virtual spring-damper systems to drive a humanoid robot's movement. The method relies on a statistical description of the springs attractor points acting in different candidate frames of reference. It extends dynamic movement primitives models by formulating the dynamical systems parameters estimation problem as a Gaussian mixture regression problem with projection in different coordinate systems. The robot exploits local variability information extracted from multiple demonstrations of movements to determine which frames are relevant for the task, and how the movement should be modulated with respect to these frames. The approach is tested on the new prototype of the COMAN compliant humanoid with time-based and time-invariant movements, including bimanual coordination skills.

UR - http://www.scopus.com/inward/record.url?scp=84891139169&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84891139169&partnerID=8YFLogxK

U2 - 10.1109/HUMANOIDS.2012.6651539

DO - 10.1109/HUMANOIDS.2012.6651539

M3 - Conference contribution

AN - SCOPUS:84891139169

SN - 9781467313698

SP - 323

EP - 329

BT - IEEE-RAS International Conference on Humanoid Robots

ER -