Learning from demonstration with partially observable task parameters using dynamic movement primitives and Gaussian process regression

Tohid Alizadeh, Milad Malekzadeh, Soheila Barzegari

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

6 Citations (Scopus)

Abstract

The problem of learning from demonstration in the case of partially observable external task parameters is addressed in this paper. Such a situation could be present in the daily life scenarios, where information regarding some task parameters are missing or partially available. In the first step, one dynamic movement primitives (DMP) model is learned for each demonstration trajectory. The parameters of the learned DMP model are recorded together with the corresponding external task parameters (query points), to create a database. Then Gaussian process regression (GPR) is used to create a model for the external task parameters and the DMP parameters. During reproduction, DMP parameters are retrieved by providing the new external task parameters and are used to regenerate the trajectory. It is shown how the learning approach could be adapted to deal with the partially observable external task parameters and regenerate the proper trajectory. The proposed methodology is applied to learn a via-point passing experiment with a lightweight robot manipulator (KUKA robot) to illustrate the efficacy of the proposed approach.

Original languageEnglish
Title of host publication2016 IEEE International Conference on Advanced Intelligent Mechatronics, AIM 2016
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages889-894
Number of pages6
Volume2016-September
ISBN (Electronic)9781509020652
DOIs
Publication statusPublished - Sep 26 2016
Event2016 IEEE International Conference on Advanced Intelligent Mechatronics, AIM 2016 - Banff, Canada
Duration: Jul 12 2016Jul 15 2016

Conference

Conference2016 IEEE International Conference on Advanced Intelligent Mechatronics, AIM 2016
CountryCanada
CityBanff
Period7/12/167/15/16

Fingerprint

Demonstrations
Trajectories
Robots
Manipulators
Experiments

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Control and Systems Engineering
  • Computer Science Applications
  • Software

Cite this

Alizadeh, T., Malekzadeh, M., & Barzegari, S. (2016). Learning from demonstration with partially observable task parameters using dynamic movement primitives and Gaussian process regression. In 2016 IEEE International Conference on Advanced Intelligent Mechatronics, AIM 2016 (Vol. 2016-September, pp. 889-894). [7576881] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/AIM.2016.7576881

Learning from demonstration with partially observable task parameters using dynamic movement primitives and Gaussian process regression. / Alizadeh, Tohid; Malekzadeh, Milad; Barzegari, Soheila.

2016 IEEE International Conference on Advanced Intelligent Mechatronics, AIM 2016. Vol. 2016-September Institute of Electrical and Electronics Engineers Inc., 2016. p. 889-894 7576881.

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

Alizadeh, T, Malekzadeh, M & Barzegari, S 2016, Learning from demonstration with partially observable task parameters using dynamic movement primitives and Gaussian process regression. in 2016 IEEE International Conference on Advanced Intelligent Mechatronics, AIM 2016. vol. 2016-September, 7576881, Institute of Electrical and Electronics Engineers Inc., pp. 889-894, 2016 IEEE International Conference on Advanced Intelligent Mechatronics, AIM 2016, Banff, Canada, 7/12/16. https://doi.org/10.1109/AIM.2016.7576881
Alizadeh T, Malekzadeh M, Barzegari S. Learning from demonstration with partially observable task parameters using dynamic movement primitives and Gaussian process regression. In 2016 IEEE International Conference on Advanced Intelligent Mechatronics, AIM 2016. Vol. 2016-September. Institute of Electrical and Electronics Engineers Inc. 2016. p. 889-894. 7576881 https://doi.org/10.1109/AIM.2016.7576881
Alizadeh, Tohid ; Malekzadeh, Milad ; Barzegari, Soheila. / Learning from demonstration with partially observable task parameters using dynamic movement primitives and Gaussian process regression. 2016 IEEE International Conference on Advanced Intelligent Mechatronics, AIM 2016. Vol. 2016-September Institute of Electrical and Electronics Engineers Inc., 2016. pp. 889-894
@inproceedings{8e6a6541fe0e4c5884b88e4dde4e6007,
title = "Learning from demonstration with partially observable task parameters using dynamic movement primitives and Gaussian process regression",
abstract = "The problem of learning from demonstration in the case of partially observable external task parameters is addressed in this paper. Such a situation could be present in the daily life scenarios, where information regarding some task parameters are missing or partially available. In the first step, one dynamic movement primitives (DMP) model is learned for each demonstration trajectory. The parameters of the learned DMP model are recorded together with the corresponding external task parameters (query points), to create a database. Then Gaussian process regression (GPR) is used to create a model for the external task parameters and the DMP parameters. During reproduction, DMP parameters are retrieved by providing the new external task parameters and are used to regenerate the trajectory. It is shown how the learning approach could be adapted to deal with the partially observable external task parameters and regenerate the proper trajectory. The proposed methodology is applied to learn a via-point passing experiment with a lightweight robot manipulator (KUKA robot) to illustrate the efficacy of the proposed approach.",
author = "Tohid Alizadeh and Milad Malekzadeh and Soheila Barzegari",
year = "2016",
month = "9",
day = "26",
doi = "10.1109/AIM.2016.7576881",
language = "English",
volume = "2016-September",
pages = "889--894",
booktitle = "2016 IEEE International Conference on Advanced Intelligent Mechatronics, AIM 2016",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
address = "United States",

}

TY - GEN

T1 - Learning from demonstration with partially observable task parameters using dynamic movement primitives and Gaussian process regression

AU - Alizadeh, Tohid

AU - Malekzadeh, Milad

AU - Barzegari, Soheila

PY - 2016/9/26

Y1 - 2016/9/26

N2 - The problem of learning from demonstration in the case of partially observable external task parameters is addressed in this paper. Such a situation could be present in the daily life scenarios, where information regarding some task parameters are missing or partially available. In the first step, one dynamic movement primitives (DMP) model is learned for each demonstration trajectory. The parameters of the learned DMP model are recorded together with the corresponding external task parameters (query points), to create a database. Then Gaussian process regression (GPR) is used to create a model for the external task parameters and the DMP parameters. During reproduction, DMP parameters are retrieved by providing the new external task parameters and are used to regenerate the trajectory. It is shown how the learning approach could be adapted to deal with the partially observable external task parameters and regenerate the proper trajectory. The proposed methodology is applied to learn a via-point passing experiment with a lightweight robot manipulator (KUKA robot) to illustrate the efficacy of the proposed approach.

AB - The problem of learning from demonstration in the case of partially observable external task parameters is addressed in this paper. Such a situation could be present in the daily life scenarios, where information regarding some task parameters are missing or partially available. In the first step, one dynamic movement primitives (DMP) model is learned for each demonstration trajectory. The parameters of the learned DMP model are recorded together with the corresponding external task parameters (query points), to create a database. Then Gaussian process regression (GPR) is used to create a model for the external task parameters and the DMP parameters. During reproduction, DMP parameters are retrieved by providing the new external task parameters and are used to regenerate the trajectory. It is shown how the learning approach could be adapted to deal with the partially observable external task parameters and regenerate the proper trajectory. The proposed methodology is applied to learn a via-point passing experiment with a lightweight robot manipulator (KUKA robot) to illustrate the efficacy of the proposed approach.

UR - http://www.scopus.com/inward/record.url?scp=84992347130&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84992347130&partnerID=8YFLogxK

U2 - 10.1109/AIM.2016.7576881

DO - 10.1109/AIM.2016.7576881

M3 - Conference contribution

AN - SCOPUS:84992347130

VL - 2016-September

SP - 889

EP - 894

BT - 2016 IEEE International Conference on Advanced Intelligent Mechatronics, AIM 2016

PB - Institute of Electrical and Electronics Engineers Inc.

ER -