Shaping Imbalance into Balance: Active Robot Guidance of Human Teachers for Better Learning from Demonstrations

Muhan Hou; Koen Hindriks; A. E. Eiben; Kim Baraka

doi:10.1109/RO-MAN57019.2023.10309481

Shaping Imbalance into Balance: Active Robot Guidance of Human Teachers for Better Learning from Demonstrations

Muhan Hou^*, Koen Hindriks, A. E. Eiben, Kim Baraka

^*Corresponding author for this work

Research output: Chapter in Book / Report / Conference proceeding › Conference contribution › Academic › peer-review

12 Downloads (Pure)

Abstract

Learning from Demonstrations (LfD) transfers skills from human teachers to robots. However, data imbalance in demonstrations can bias policies towards majority situations. Previous work attempted to solve this problem after data collection, but few efforts were made to maintain a balanced distribution from the phase of data acquisition. Our method accounts for the influence of robots on human teachers and enables robots to actively guide interaction to approximate demonstration distributions to target distributions. Simulated and real-world experiments validated the method's efficacy in shaping demonstration distribution into various target distributions and robustness to various levels of uncertainties. Also, our method significantly improved the generalization ability of robot learning when LfD policies were trained with data collected by our method compared to natural data collection.

Original language	English
Title of host publication	2023 32nd IEEE International Conference on Robot and Human Interactive Communication (RO-MAN)
Subtitle of host publication	[Proceedings]
Publisher	IEEE Computer Society
Pages	1737-1744
Number of pages	8
ISBN (Electronic)	9798350336702
ISBN (Print)	9798350336719
DOIs	https://doi.org/10.1109/RO-MAN57019.2023.10309481
Publication status	Published - 2023
Event	32nd IEEE International Conference on Robot and Human Interactive Communication, RO-MAN 2023 - Busan, Korea, Republic of Duration: 28 Aug 2023 → 31 Aug 2023

Publication series

Name	IEEE International Workshop on Robot and Human Communication, RO-MAN
ISSN (Print)	1944-9445
ISSN (Electronic)	1944-9437

Conference

Conference	32nd IEEE International Conference on Robot and Human Interactive Communication, RO-MAN 2023
Country/Territory	Korea, Republic of
City	Busan
Period	28/08/23 → 31/08/23

Bibliographical note

Publisher Copyright:
© 2023 IEEE.

Keywords

data collection
data imbalance
human-robot interaction
learning from demonstration

Access to Document

10.1109/RO-MAN57019.2023.10309481

Shaping_Imbalance_into_Balance_Active_Robot_Guidance_of_Human_Teachers_for_Better_Learning_from_DemonstrationsFinal published version, 3.05 MBLicence: Article 25fa Dutch Copyright Act

Persistent URL (handle)

Persistent link

Cite this

Hou, M., Hindriks, K., Eiben, A. E., & Baraka, K. (2023). Shaping Imbalance into Balance: Active Robot Guidance of Human Teachers for Better Learning from Demonstrations. In 2023 32nd IEEE International Conference on Robot and Human Interactive Communication (RO-MAN): [Proceedings] (pp. 1737-1744). (IEEE International Workshop on Robot and Human Communication, RO-MAN). IEEE Computer Society. https://doi.org/10.1109/RO-MAN57019.2023.10309481

Hou, Muhan ; Hindriks, Koen ; Eiben, A. E. et al. / Shaping Imbalance into Balance : Active Robot Guidance of Human Teachers for Better Learning from Demonstrations. 2023 32nd IEEE International Conference on Robot and Human Interactive Communication (RO-MAN): [Proceedings]. IEEE Computer Society, 2023. pp. 1737-1744 (IEEE International Workshop on Robot and Human Communication, RO-MAN).

@inproceedings{df63fe28f0484519b3aeed3f4d3d3bd5,

title = "Shaping Imbalance into Balance: Active Robot Guidance of Human Teachers for Better Learning from Demonstrations",

abstract = "Learning from Demonstrations (LfD) transfers skills from human teachers to robots. However, data imbalance in demonstrations can bias policies towards majority situations. Previous work attempted to solve this problem after data collection, but few efforts were made to maintain a balanced distribution from the phase of data acquisition. Our method accounts for the influence of robots on human teachers and enables robots to actively guide interaction to approximate demonstration distributions to target distributions. Simulated and real-world experiments validated the method's efficacy in shaping demonstration distribution into various target distributions and robustness to various levels of uncertainties. Also, our method significantly improved the generalization ability of robot learning when LfD policies were trained with data collected by our method compared to natural data collection.",

keywords = "data collection, data imbalance, human-robot interaction, learning from demonstration",

author = "Muhan Hou and Koen Hindriks and Eiben, {A. E.} and Kim Baraka",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 32nd IEEE International Conference on Robot and Human Interactive Communication, RO-MAN 2023 ; Conference date: 28-08-2023 Through 31-08-2023",

year = "2023",

doi = "10.1109/RO-MAN57019.2023.10309481",

language = "English",

isbn = "9798350336719",

series = "IEEE International Workshop on Robot and Human Communication, RO-MAN",

publisher = "IEEE Computer Society",

pages = "1737--1744",

booktitle = "2023 32nd IEEE International Conference on Robot and Human Interactive Communication (RO-MAN)",

address = "United States",

}

Hou, M , Hindriks, K , Eiben, AE & Baraka, K 2023, Shaping Imbalance into Balance: Active Robot Guidance of Human Teachers for Better Learning from Demonstrations. in 2023 32nd IEEE International Conference on Robot and Human Interactive Communication (RO-MAN): [Proceedings]. IEEE International Workshop on Robot and Human Communication, RO-MAN, IEEE Computer Society, pp. 1737-1744, 32nd IEEE International Conference on Robot and Human Interactive Communication, RO-MAN 2023, Busan, Korea, Republic of, 28/08/23. https://doi.org/10.1109/RO-MAN57019.2023.10309481

Shaping Imbalance into Balance: Active Robot Guidance of Human Teachers for Better Learning from Demonstrations. / Hou, Muhan ; Hindriks, Koen ; Eiben, A. E. et al.
2023 32nd IEEE International Conference on Robot and Human Interactive Communication (RO-MAN): [Proceedings]. IEEE Computer Society, 2023. p. 1737-1744 (IEEE International Workshop on Robot and Human Communication, RO-MAN).

Research output: Chapter in Book / Report / Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Shaping Imbalance into Balance

T2 - 32nd IEEE International Conference on Robot and Human Interactive Communication, RO-MAN 2023

AU - Hou, Muhan

AU - Hindriks, Koen

AU - Eiben, A. E.

AU - Baraka, Kim

PY - 2023

Y1 - 2023

N2 - Learning from Demonstrations (LfD) transfers skills from human teachers to robots. However, data imbalance in demonstrations can bias policies towards majority situations. Previous work attempted to solve this problem after data collection, but few efforts were made to maintain a balanced distribution from the phase of data acquisition. Our method accounts for the influence of robots on human teachers and enables robots to actively guide interaction to approximate demonstration distributions to target distributions. Simulated and real-world experiments validated the method's efficacy in shaping demonstration distribution into various target distributions and robustness to various levels of uncertainties. Also, our method significantly improved the generalization ability of robot learning when LfD policies were trained with data collected by our method compared to natural data collection.

AB - Learning from Demonstrations (LfD) transfers skills from human teachers to robots. However, data imbalance in demonstrations can bias policies towards majority situations. Previous work attempted to solve this problem after data collection, but few efforts were made to maintain a balanced distribution from the phase of data acquisition. Our method accounts for the influence of robots on human teachers and enables robots to actively guide interaction to approximate demonstration distributions to target distributions. Simulated and real-world experiments validated the method's efficacy in shaping demonstration distribution into various target distributions and robustness to various levels of uncertainties. Also, our method significantly improved the generalization ability of robot learning when LfD policies were trained with data collected by our method compared to natural data collection.

KW - data collection

KW - data imbalance

KW - human-robot interaction

KW - learning from demonstration

UR - http://www.scopus.com/inward/record.url?scp=85186993381&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85186993381&partnerID=8YFLogxK

U2 - 10.1109/RO-MAN57019.2023.10309481

DO - 10.1109/RO-MAN57019.2023.10309481

M3 - Conference contribution

AN - SCOPUS:85186993381

SN - 9798350336719

T3 - IEEE International Workshop on Robot and Human Communication, RO-MAN

SP - 1737

EP - 1744

BT - 2023 32nd IEEE International Conference on Robot and Human Interactive Communication (RO-MAN)

PB - IEEE Computer Society

Y2 - 28 August 2023 through 31 August 2023

ER -

Hou M , Hindriks K , Eiben AE , Baraka K. Shaping Imbalance into Balance: Active Robot Guidance of Human Teachers for Better Learning from Demonstrations. In 2023 32nd IEEE International Conference on Robot and Human Interactive Communication (RO-MAN): [Proceedings]. IEEE Computer Society. 2023. p. 1737-1744. (IEEE International Workshop on Robot and Human Communication, RO-MAN). Epub 2023 Nov 13. doi: 10.1109/RO-MAN57019.2023.10309481

Shaping Imbalance into Balance: Active Robot Guidance of Human Teachers for Better Learning from Demonstrations

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Persistent URL (handle)

Other files and links

Fingerprint

Cite this