Safe Robot Learning in Assistive Devices through Neural Network Repair

Safe Robot Learning in Assistive Devices through Neural Network Repair

PMLR 205:2148-2158, 2023 Assistive robotic devices are a particularly promising field of application for neural networks (NN) due to the need for personalization and hard-to-model human-machine interaction dynamics. However, NN based estimators and controllers may produce potentially unsafe outputs...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Majd, Keyvan, Clark, Geoffrey, Khandait, Tanmay, Zhou, Siyu, Sankaranarayanan, Sriram, Fainekos, Georgios, Amor, Heni Ben
Format: Artikel
Sprache:eng
Schlagworte:
Computer Science - Robotics
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Majd, Keyvan
Clark, Geoffrey
Khandait, Tanmay
Zhou, Siyu
Sankaranarayanan, Sriram
Fainekos, Georgios
Amor, Heni Ben
description PMLR 205:2148-2158, 2023 Assistive robotic devices are a particularly promising field of application for neural networks (NN) due to the need for personalization and hard-to-model human-machine interaction dynamics. However, NN based estimators and controllers may produce potentially unsafe outputs over previously unseen data points. In this paper, we introduce an algorithm for updating NN control policies to satisfy a given set of formal safety constraints, while also optimizing the original loss function. Given a set of mixed-integer linear constraints, we define the NN repair problem as a Mixed Integer Quadratic Program (MIQP). In extensive experiments, we demonstrate the efficacy of our repair method in generating safe policies for a lower-leg prosthesis.
doi_str_mv 10.48550/arxiv.2303.04431
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2303_04431</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2303_04431</sourcerecordid><originalsourceid>FETCH-LOGICAL-a671-d0e4d7e66922a179c7d6656fe3b449afe07aab83300aacf7fe0bb1200dc2d4343</originalsourceid><addsrcrecordid>eNotj71OwzAURr10QC0PwIRfIOH6J3YzVoUCUgRS6R5dxzetRUkqOw3w9oTCdKRv-HQOYzcCcr0sCrjD-BXGXCpQOWitxBXbvGFLfNu7fuAVYexCt-eh46uUQhrCSPyextBQ4sMh9uf9gb_QOeJxwvDZx3e-pROGuGCzFo-Jrv85Z7vNw279lFWvj8_rVZWhsSLzQNpbMqaUEoUtG-uNKUxLymldTiJgEd1SKQDEprXT4JyQAL6RXiut5uz27_YSUp9i-MD4Xf8G1Zcg9QPX5UYO</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Safe Robot Learning in Assistive Devices through Neural Network Repair</title><source>arXiv.org</source><creator>Majd, Keyvan ; Clark, Geoffrey ; Khandait, Tanmay ; Zhou, Siyu ; Sankaranarayanan, Sriram ; Fainekos, Georgios ; Amor, Heni Ben</creator><creatorcontrib>Majd, Keyvan ; Clark, Geoffrey ; Khandait, Tanmay ; Zhou, Siyu ; Sankaranarayanan, Sriram ; Fainekos, Georgios ; Amor, Heni Ben</creatorcontrib><description>PMLR 205:2148-2158, 2023 Assistive robotic devices are a particularly promising field of application for neural networks (NN) due to the need for personalization and hard-to-model human-machine interaction dynamics. However, NN based estimators and controllers may produce potentially unsafe outputs over previously unseen data points. In this paper, we introduce an algorithm for updating NN control policies to satisfy a given set of formal safety constraints, while also optimizing the original loss function. Given a set of mixed-integer linear constraints, we define the NN repair problem as a Mixed Integer Quadratic Program (MIQP). In extensive experiments, we demonstrate the efficacy of our repair method in generating safe policies for a lower-leg prosthesis.</description><identifier>DOI: 10.48550/arxiv.2303.04431</identifier><language>eng</language><subject>Computer Science - Robotics</subject><creationdate>2023-03</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2303.04431$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2303.04431$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Majd, Keyvan</creatorcontrib><creatorcontrib>Clark, Geoffrey</creatorcontrib><creatorcontrib>Khandait, Tanmay</creatorcontrib><creatorcontrib>Zhou, Siyu</creatorcontrib><creatorcontrib>Sankaranarayanan, Sriram</creatorcontrib><creatorcontrib>Fainekos, Georgios</creatorcontrib><creatorcontrib>Amor, Heni Ben</creatorcontrib><title>Safe Robot Learning in Assistive Devices through Neural Network Repair</title><description>PMLR 205:2148-2158, 2023 Assistive robotic devices are a particularly promising field of application for neural networks (NN) due to the need for personalization and hard-to-model human-machine interaction dynamics. However, NN based estimators and controllers may produce potentially unsafe outputs over previously unseen data points. In this paper, we introduce an algorithm for updating NN control policies to satisfy a given set of formal safety constraints, while also optimizing the original loss function. Given a set of mixed-integer linear constraints, we define the NN repair problem as a Mixed Integer Quadratic Program (MIQP). In extensive experiments, we demonstrate the efficacy of our repair method in generating safe policies for a lower-leg prosthesis.</description><subject>Computer Science - Robotics</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotj71OwzAURr10QC0PwIRfIOH6J3YzVoUCUgRS6R5dxzetRUkqOw3w9oTCdKRv-HQOYzcCcr0sCrjD-BXGXCpQOWitxBXbvGFLfNu7fuAVYexCt-eh46uUQhrCSPyextBQ4sMh9uf9gb_QOeJxwvDZx3e-pROGuGCzFo-Jrv85Z7vNw279lFWvj8_rVZWhsSLzQNpbMqaUEoUtG-uNKUxLymldTiJgEd1SKQDEprXT4JyQAL6RXiut5uz27_YSUp9i-MD4Xf8G1Zcg9QPX5UYO</recordid><startdate>20230308</startdate><enddate>20230308</enddate><creator>Majd, Keyvan</creator><creator>Clark, Geoffrey</creator><creator>Khandait, Tanmay</creator><creator>Zhou, Siyu</creator><creator>Sankaranarayanan, Sriram</creator><creator>Fainekos, Georgios</creator><creator>Amor, Heni Ben</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20230308</creationdate><title>Safe Robot Learning in Assistive Devices through Neural Network Repair</title><author>Majd, Keyvan ; Clark, Geoffrey ; Khandait, Tanmay ; Zhou, Siyu ; Sankaranarayanan, Sriram ; Fainekos, Georgios ; Amor, Heni Ben</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a671-d0e4d7e66922a179c7d6656fe3b449afe07aab83300aacf7fe0bb1200dc2d4343</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Computer Science - Robotics</topic><toplevel>online_resources</toplevel><creatorcontrib>Majd, Keyvan</creatorcontrib><creatorcontrib>Clark, Geoffrey</creatorcontrib><creatorcontrib>Khandait, Tanmay</creatorcontrib><creatorcontrib>Zhou, Siyu</creatorcontrib><creatorcontrib>Sankaranarayanan, Sriram</creatorcontrib><creatorcontrib>Fainekos, Georgios</creatorcontrib><creatorcontrib>Amor, Heni Ben</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Majd, Keyvan</au><au>Clark, Geoffrey</au><au>Khandait, Tanmay</au><au>Zhou, Siyu</au><au>Sankaranarayanan, Sriram</au><au>Fainekos, Georgios</au><au>Amor, Heni Ben</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Safe Robot Learning in Assistive Devices through Neural Network Repair</atitle><date>2023-03-08</date><risdate>2023</risdate><abstract>PMLR 205:2148-2158, 2023 Assistive robotic devices are a particularly promising field of application for neural networks (NN) due to the need for personalization and hard-to-model human-machine interaction dynamics. However, NN based estimators and controllers may produce potentially unsafe outputs over previously unseen data points. In this paper, we introduce an algorithm for updating NN control policies to satisfy a given set of formal safety constraints, while also optimizing the original loss function. Given a set of mixed-integer linear constraints, we define the NN repair problem as a Mixed Integer Quadratic Program (MIQP). In extensive experiments, we demonstrate the efficacy of our repair method in generating safe policies for a lower-leg prosthesis.</abstract><doi>10.48550/arxiv.2303.04431</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2303.04431
ispartof
issn
language eng
recordid cdi_arxiv_primary_2303_04431
source arXiv.org
subjects Computer Science - Robotics
title Safe Robot Learning in Assistive Devices through Neural Network Repair
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-25T23%3A28%3A29IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Safe%20Robot%20Learning%20in%20Assistive%20Devices%20through%20Neural%20Network%20Repair&rft.au=Majd,%20Keyvan&rft.date=2023-03-08&rft_id=info:doi/10.48550/arxiv.2303.04431&rft_dat=%3Carxiv_GOX%3E2303_04431%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true