Human-robot interaction assessment using dynamic engagement profiles

Alin Drimus, Nicole Poltorak

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

Abstract

This paper addresses the use of convolutional neural networks for image analysis resulting in an engagement metric that can be used to assess the quality of human robot interactions. We propose a method based on a pretrained convolutional network able to map emotions onto a continuous [0-1] interval, where 0 represents disengaged and 1 fully engaged. The network shows a good accuracy at recognizing the engagement state of humans given positive emotions. A time based analysis of interaction experiments between small humanoid robots and humans provides time series of engagement estimates, which are further used to understand the nature of the interaction as well as the overall mood and interest of the participant during the experiment. The method allows a real-time implementation and supports a quantitative and qualitative assessment of a human robot interaction with respect to a positive engagement and is applicable to humanoid robotics as well as other related contexts.
Original languageEnglish
Title of host publicationProceedings of the 17th International Conference on Humanoid Robotics
PublisherIEEE
Publication date2017
Pages649-654
ISBN (Print)978-1-5386-4679-3
ISBN (Electronic)978-1-5386-4678-6, 978-1-5386-4677-9
DOIs
Publication statusPublished - 2017
Event2017 IEEE-RAS 17th International Conference on Humanoid Robotics - Birmingham, United Kingdom
Duration: 15 Nov 201717 Nov 2017
http://humanoids2017.loria.fr/

Conference

Conference2017 IEEE-RAS 17th International Conference on Humanoid Robotics
CountryUnited Kingdom
CityBirmingham
Period15/11/201717/11/2017
Internet address

Fingerprint

Human robot interaction
Image analysis
Time series
Robotics
Experiments
Robots
Neural networks

Keywords

  • convolutional neural networks
  • engagement
  • human-robot-interaction

Cite this

Drimus, A., & Poltorak, N. (2017). Human-robot interaction assessment using dynamic engagement profiles. In Proceedings of the 17th International Conference on Humanoid Robotics (pp. 649-654). IEEE. https://doi.org/10.1109/HUMANOIDS.2017.8246941
Drimus, Alin ; Poltorak, Nicole. / Human-robot interaction assessment using dynamic engagement profiles. Proceedings of the 17th International Conference on Humanoid Robotics. IEEE, 2017. pp. 649-654
@inproceedings{0d22c57691af4d1dae5c62b6ad9ef653,
title = "Human-robot interaction assessment using dynamic engagement profiles",
abstract = "This paper addresses the use of convolutional neural networks for image analysis resulting in an engagement metric that can be used to assess the quality of human robot interactions. We propose a method based on a pretrained convolutional network able to map emotions onto a continuous [0-1] interval, where 0 represents disengaged and 1 fully engaged. The network shows a good accuracy at recognizing the engagement state of humans given positive emotions. A time based analysis of interaction experiments between small humanoid robots and humans provides time series of engagement estimates, which are further used to understand the nature of the interaction as well as the overall mood and interest of the participant during the experiment. The method allows a real-time implementation and supports a quantitative and qualitative assessment of a human robot interaction with respect to a positive engagement and is applicable to humanoid robotics as well as other related contexts.",
keywords = "convolutional neural networks, engagement, human-robot-interaction",
author = "Alin Drimus and Nicole Poltorak",
year = "2017",
doi = "10.1109/HUMANOIDS.2017.8246941",
language = "English",
isbn = "978-1-5386-4679-3",
pages = "649--654",
booktitle = "Proceedings of the 17th International Conference on Humanoid Robotics",
publisher = "IEEE",
address = "United States",

}

Drimus, A & Poltorak, N 2017, Human-robot interaction assessment using dynamic engagement profiles. in Proceedings of the 17th International Conference on Humanoid Robotics. IEEE, pp. 649-654, 2017 IEEE-RAS 17th International Conference on Humanoid Robotics, Birmingham, United Kingdom, 15/11/2017. https://doi.org/10.1109/HUMANOIDS.2017.8246941

Human-robot interaction assessment using dynamic engagement profiles. / Drimus, Alin; Poltorak, Nicole.

Proceedings of the 17th International Conference on Humanoid Robotics. IEEE, 2017. p. 649-654.

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

TY - GEN

T1 - Human-robot interaction assessment using dynamic engagement profiles

AU - Drimus, Alin

AU - Poltorak, Nicole

PY - 2017

Y1 - 2017

N2 - This paper addresses the use of convolutional neural networks for image analysis resulting in an engagement metric that can be used to assess the quality of human robot interactions. We propose a method based on a pretrained convolutional network able to map emotions onto a continuous [0-1] interval, where 0 represents disengaged and 1 fully engaged. The network shows a good accuracy at recognizing the engagement state of humans given positive emotions. A time based analysis of interaction experiments between small humanoid robots and humans provides time series of engagement estimates, which are further used to understand the nature of the interaction as well as the overall mood and interest of the participant during the experiment. The method allows a real-time implementation and supports a quantitative and qualitative assessment of a human robot interaction with respect to a positive engagement and is applicable to humanoid robotics as well as other related contexts.

AB - This paper addresses the use of convolutional neural networks for image analysis resulting in an engagement metric that can be used to assess the quality of human robot interactions. We propose a method based on a pretrained convolutional network able to map emotions onto a continuous [0-1] interval, where 0 represents disengaged and 1 fully engaged. The network shows a good accuracy at recognizing the engagement state of humans given positive emotions. A time based analysis of interaction experiments between small humanoid robots and humans provides time series of engagement estimates, which are further used to understand the nature of the interaction as well as the overall mood and interest of the participant during the experiment. The method allows a real-time implementation and supports a quantitative and qualitative assessment of a human robot interaction with respect to a positive engagement and is applicable to humanoid robotics as well as other related contexts.

KW - convolutional neural networks

KW - engagement

KW - human-robot-interaction

U2 - 10.1109/HUMANOIDS.2017.8246941

DO - 10.1109/HUMANOIDS.2017.8246941

M3 - Article in proceedings

SN - 978-1-5386-4679-3

SP - 649

EP - 654

BT - Proceedings of the 17th International Conference on Humanoid Robotics

PB - IEEE

ER -

Drimus A, Poltorak N. Human-robot interaction assessment using dynamic engagement profiles. In Proceedings of the 17th International Conference on Humanoid Robotics. IEEE. 2017. p. 649-654 https://doi.org/10.1109/HUMANOIDS.2017.8246941