Learning robot anomaly recovery skills from multiple time-driven demonstrations

Wu, Hongmin; Yan, Wu; Xu, Zhihao; Li, Shuai; Zhou, Xuefeng

doi:10.1016/j.neucom.2021.08.036

Journal article 649 views 110 downloads

Learning robot anomaly recovery skills from multiple time-driven demonstrations

Hongmin Wu, Wu Yan, Zhihao Xu, Shuai Li

, Xuefeng Zhou

Neurocomputing, Volume: 464, Pages: 522 - 532

Swansea University Author: Shuai Li

PDF | Accepted Manuscript

©2021 All rights reserved. All article content, except where otherwise noted, is licensed under a Creative Commons Attribution Non-Commercial No Derivatives License (CC-BY-NC-ND)
Download (1.94MB)

Check full text

DOI (Published version): 10.1016/j.neucom.2021.08.036

Abstract

Robots are prone to making anomalies when performing manipulation tasks in unstructured environments, it is often desirable to rapidly adapt the robotic behavior to avoid environmental changes by learning from experts’ demonstrations. We propose a framework for learning robot anomaly recovery skills...

Full description

Published in:	Neurocomputing
ISSN:	0925-2312
Published:	Elsevier BV 2021
Online Access:	Check full text
URI:	https://cronfa.swan.ac.uk/Record/cronfa57655
Tags:	Add Tag No Tags, Be the first to tag this record!

first_indexed	2021-08-19T08:49:47Z
last_indexed	2022-02-15T04:25:32Z
id	cronfa57655
recordtype	SURis
fullrecord	<?xml version="1.0"?><rfc1807><datestamp>2022-02-14T13:57:18.8632215</datestamp><bib-version>v2</bib-version><id>57655</id><entry>2021-08-19</entry><title>Learning robot anomaly recovery skills from multiple time-driven demonstrations</title><swanseaauthors><author><sid>42ff9eed09bcd109fbbe484a0f99a8a8</sid><ORCID>0000-0001-8316-5289</ORCID><firstname>Shuai</firstname><surname>Li</surname><name>Shuai Li</name><active>true</active><ethesisStudent>false</ethesisStudent></author></swanseaauthors><date>2021-08-19</date><deptcode>MECH</deptcode><abstract>Robots are prone to making anomalies when performing manipulation tasks in unstructured environments, it is often desirable to rapidly adapt the robotic behavior to avoid environmental changes by learning from experts’ demonstrations. We propose a framework for learning robot anomaly recovery skills from time-driven demonstrations based on a Gaussian process regression with prior mean derived by Gaussian mixture regression, named as mean-prior GPR (MP-GPR), which allows an end-user to adjust the anomalous trajectory intuitively by simultaneously considering the variability of the demonstrations and the adaptation of recovery skills. Evaluations are divided into two phases, a benchmarking dataset with robot reaching, pushing, writing, and pressing tasks are first used to verify the path accuracy and variability, and then a real-time robot bin-picking task for evaluating the adaptation of the framework. Our method has a fair comparison with probabilistic-based methods in the field of robot learning from demonstrations, including Gaussian mixture regression, probabilistic movement primitives, and kernelized movement primitives. The results indicate that our proposed method can efficiently encode the variability from multiple demonstrations and rapidly anomaly recovery skills learning by modulating a learned trajectory to safe via-points.</abstract><type>Journal Article</type><journal>Neurocomputing</journal><volume>464</volume><journalNumber/><paginationStart>522</paginationStart><paginationEnd>532</paginationEnd><publisher>Elsevier BV</publisher><placeOfPublication/><isbnPrint/><isbnElectronic/><issnPrint>0925-2312</issnPrint><issnElectronic/><keywords>Robot skill learning; Gaussian process; Robot recovery</keywords><publishedDay>13</publishedDay><publishedMonth>11</publishedMonth><publishedYear>2021</publishedYear><publishedDate>2021-11-13</publishedDate><doi>10.1016/j.neucom.2021.08.036</doi><url/><notes/><college>COLLEGE NANME</college><department>Mechanical Engineering</department><CollegeCode>COLLEGE CODE</CollegeCode><DepartmentCode>MECH</DepartmentCode><institution>Swansea University</institution><apcterm/><lastEdited>2022-02-14T13:57:18.8632215</lastEdited><Created>2021-08-19T09:47:57.3760413</Created><path><level id="1">Faculty of Science and Engineering</level><level id="2">School of Aerospace, Civil, Electrical, General and Mechanical Engineering - Mechanical Engineering</level></path><authors><author><firstname>Hongmin</firstname><surname>Wu</surname><order>1</order></author><author><firstname>Wu</firstname><surname>Yan</surname><order>2</order></author><author><firstname>Zhihao</firstname><surname>Xu</surname><order>3</order></author><author><firstname>Shuai</firstname><surname>Li</surname><orcid>0000-0001-8316-5289</orcid><order>4</order></author><author><firstname>Xuefeng</firstname><surname>Zhou</surname><order>5</order></author></authors><documents><document><filename>57655__20674__080d394762514446becb2ae86e0d38d0.pdf</filename><originalFilename>57655.pdf</originalFilename><uploaded>2021-08-19T09:49:26.9162479</uploaded><type>Output</type><contentLength>2034719</contentLength><contentType>application/pdf</contentType><version>Accepted Manuscript</version><cronfaStatus>true</cronfaStatus><embargoDate>2022-08-14T00:00:00.0000000</embargoDate><documentNotes>©2021 All rights reserved. All article content, except where otherwise noted, is licensed under a Creative Commons Attribution Non-Commercial No Derivatives License (CC-BY-NC-ND)</documentNotes><copyrightCorrect>true</copyrightCorrect><language>eng</language><licence>https://creativecommons.org/licenses/by-nc-nd/4.0/</licence></document></documents><OutputDurs/></rfc1807>
spelling	2022-02-14T13:57:18.8632215 v2 57655 2021-08-19 Learning robot anomaly recovery skills from multiple time-driven demonstrations 42ff9eed09bcd109fbbe484a0f99a8a8 0000-0001-8316-5289 Shuai Li Shuai Li true false 2021-08-19 MECH Robots are prone to making anomalies when performing manipulation tasks in unstructured environments, it is often desirable to rapidly adapt the robotic behavior to avoid environmental changes by learning from experts’ demonstrations. We propose a framework for learning robot anomaly recovery skills from time-driven demonstrations based on a Gaussian process regression with prior mean derived by Gaussian mixture regression, named as mean-prior GPR (MP-GPR), which allows an end-user to adjust the anomalous trajectory intuitively by simultaneously considering the variability of the demonstrations and the adaptation of recovery skills. Evaluations are divided into two phases, a benchmarking dataset with robot reaching, pushing, writing, and pressing tasks are first used to verify the path accuracy and variability, and then a real-time robot bin-picking task for evaluating the adaptation of the framework. Our method has a fair comparison with probabilistic-based methods in the field of robot learning from demonstrations, including Gaussian mixture regression, probabilistic movement primitives, and kernelized movement primitives. The results indicate that our proposed method can efficiently encode the variability from multiple demonstrations and rapidly anomaly recovery skills learning by modulating a learned trajectory to safe via-points. Journal Article Neurocomputing 464 522 532 Elsevier BV 0925-2312 Robot skill learning; Gaussian process; Robot recovery 13 11 2021 2021-11-13 10.1016/j.neucom.2021.08.036 COLLEGE NANME Mechanical Engineering COLLEGE CODE MECH Swansea University 2022-02-14T13:57:18.8632215 2021-08-19T09:47:57.3760413 Faculty of Science and Engineering School of Aerospace, Civil, Electrical, General and Mechanical Engineering - Mechanical Engineering Hongmin Wu 1 Wu Yan 2 Zhihao Xu 3 Shuai Li 0000-0001-8316-5289 4 Xuefeng Zhou 5 57655__20674__080d394762514446becb2ae86e0d38d0.pdf 57655.pdf 2021-08-19T09:49:26.9162479 Output 2034719 application/pdf Accepted Manuscript true 2022-08-14T00:00:00.0000000 ©2021 All rights reserved. All article content, except where otherwise noted, is licensed under a Creative Commons Attribution Non-Commercial No Derivatives License (CC-BY-NC-ND) true eng https://creativecommons.org/licenses/by-nc-nd/4.0/
title	Learning robot anomaly recovery skills from multiple time-driven demonstrations
spellingShingle	Learning robot anomaly recovery skills from multiple time-driven demonstrations Shuai Li
title_short	Learning robot anomaly recovery skills from multiple time-driven demonstrations
title_full	Learning robot anomaly recovery skills from multiple time-driven demonstrations
title_fullStr	Learning robot anomaly recovery skills from multiple time-driven demonstrations
title_full_unstemmed	Learning robot anomaly recovery skills from multiple time-driven demonstrations
title_sort	Learning robot anomaly recovery skills from multiple time-driven demonstrations
author_id_str_mv	42ff9eed09bcd109fbbe484a0f99a8a8
author_id_fullname_str_mv	42ff9eed09bcd109fbbe484a0f99a8a8_***_Shuai Li
author	Shuai Li
author2	Hongmin Wu Wu Yan Zhihao Xu Shuai Li Xuefeng Zhou
format	Journal article
container_title	Neurocomputing
container_volume	464
container_start_page	522
publishDate	2021
institution	Swansea University
issn	0925-2312
doi_str_mv	10.1016/j.neucom.2021.08.036
publisher	Elsevier BV
college_str	Faculty of Science and Engineering
hierarchytype
hierarchy_top_id	facultyofscienceandengineering
hierarchy_top_title	Faculty of Science and Engineering
hierarchy_parent_id	facultyofscienceandengineering
hierarchy_parent_title	Faculty of Science and Engineering
department_str	School of Aerospace, Civil, Electrical, General and Mechanical Engineering - Mechanical Engineering{{{_:::_}}}Faculty of Science and Engineering{{{_:::_}}}School of Aerospace, Civil, Electrical, General and Mechanical Engineering - Mechanical Engineering
document_store_str	1
active_str	0
description	Robots are prone to making anomalies when performing manipulation tasks in unstructured environments, it is often desirable to rapidly adapt the robotic behavior to avoid environmental changes by learning from experts’ demonstrations. We propose a framework for learning robot anomaly recovery skills from time-driven demonstrations based on a Gaussian process regression with prior mean derived by Gaussian mixture regression, named as mean-prior GPR (MP-GPR), which allows an end-user to adjust the anomalous trajectory intuitively by simultaneously considering the variability of the demonstrations and the adaptation of recovery skills. Evaluations are divided into two phases, a benchmarking dataset with robot reaching, pushing, writing, and pressing tasks are first used to verify the path accuracy and variability, and then a real-time robot bin-picking task for evaluating the adaptation of the framework. Our method has a fair comparison with probabilistic-based methods in the field of robot learning from demonstrations, including Gaussian mixture regression, probabilistic movement primitives, and kernelized movement primitives. The results indicate that our proposed method can efficiently encode the variability from multiple demonstrations and rapidly anomaly recovery skills learning by modulating a learned trajectory to safe via-points.
published_date	2021-11-13T04:13:34Z
_version_	1763753920129138688
score	11.037056

Learning robot anomaly recovery skills from multiple time-driven demonstrations

Similar Items