Exploring the Adversarial Vulnerabilities of Vision-Language-Action Models in Robotics

Recently in robotics, Vision-Language-Action (VLA) models have emerged as a transformative approach, enabling robots to execute complex tasks by integrating visual and linguistic inputs within an end-to-end learning framework. While VLA models offer significant capabilities, they also introduce new...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2024-11
Hauptverfasser: Wang, Taowen, Liu, Dongfang, Liang, James Chenhao, Yang, Wenhao, Wang, Qifan, Cheng, Han, Luo, Jiebo, Tang, Ruixiang
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Wang, Taowen
Liu, Dongfang
Liang, James Chenhao
Yang, Wenhao
Wang, Qifan
Cheng, Han
Luo, Jiebo
Tang, Ruixiang
description Recently in robotics, Vision-Language-Action (VLA) models have emerged as a transformative approach, enabling robots to execute complex tasks by integrating visual and linguistic inputs within an end-to-end learning framework. While VLA models offer significant capabilities, they also introduce new attack surfaces, making them vulnerable to adversarial attacks. With these vulnerabilities largely unexplored, this paper systematically quantifies the robustness of VLA-based robotic systems. Recognizing the unique demands of robotic execution, our attack objectives target the inherent spatial and functional characteristics of robotic systems. In particular, we introduce an untargeted position-aware attack objective that leverages spatial foundations to destabilize robotic actions, and a targeted attack objective that manipulates the robotic trajectory. Additionally, we design an adversarial patch generation approach that places a small, colorful patch within the camera's view, effectively executing the attack in both digital and physical environments. Our evaluation reveals a marked degradation in task success rates, with up to a 100\% reduction across a suite of simulated robotic tasks, highlighting critical security gaps in current VLA architectures. By unveiling these vulnerabilities and proposing actionable evaluation metrics, this work advances both the understanding and enhancement of safety for VLA-based robotic systems, underscoring the necessity for developing robust defense strategies prior to physical-world deployments.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_3131951138</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>3131951138</sourcerecordid><originalsourceid>FETCH-proquest_journals_31319511383</originalsourceid><addsrcrecordid>eNqNikELgjAYQEcQJOV_GHQeuC3LjhJGh7pEeJWp0z4Zm-2b0c_PQz-g0-Px3oJEQkrOsp0QKxIjDkmSiP1BpKmMSFl8RuM82J6Gp6Z5-9YelQdlaDkZq72qwUAAjdR1tAQEZ9lV2X5SvWZ5E2anN9dqgxQsvbvaBWhwQ5adMqjjH9dkey4epwsbvXtNGkM1uMnbOVWSS35MOZeZ_O_6AjlyQWM</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>3131951138</pqid></control><display><type>article</type><title>Exploring the Adversarial Vulnerabilities of Vision-Language-Action Models in Robotics</title><source>Free E- Journals</source><creator>Wang, Taowen ; Liu, Dongfang ; Liang, James Chenhao ; Yang, Wenhao ; Wang, Qifan ; Cheng, Han ; Luo, Jiebo ; Tang, Ruixiang</creator><creatorcontrib>Wang, Taowen ; Liu, Dongfang ; Liang, James Chenhao ; Yang, Wenhao ; Wang, Qifan ; Cheng, Han ; Luo, Jiebo ; Tang, Ruixiang</creatorcontrib><description>Recently in robotics, Vision-Language-Action (VLA) models have emerged as a transformative approach, enabling robots to execute complex tasks by integrating visual and linguistic inputs within an end-to-end learning framework. While VLA models offer significant capabilities, they also introduce new attack surfaces, making them vulnerable to adversarial attacks. With these vulnerabilities largely unexplored, this paper systematically quantifies the robustness of VLA-based robotic systems. Recognizing the unique demands of robotic execution, our attack objectives target the inherent spatial and functional characteristics of robotic systems. In particular, we introduce an untargeted position-aware attack objective that leverages spatial foundations to destabilize robotic actions, and a targeted attack objective that manipulates the robotic trajectory. Additionally, we design an adversarial patch generation approach that places a small, colorful patch within the camera's view, effectively executing the attack in both digital and physical environments. Our evaluation reveals a marked degradation in task success rates, with up to a 100\% reduction across a suite of simulated robotic tasks, highlighting critical security gaps in current VLA architectures. By unveiling these vulnerabilities and proposing actionable evaluation metrics, this work advances both the understanding and enhancement of safety for VLA-based robotic systems, underscoring the necessity for developing robust defense strategies prior to physical-world deployments.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Robotics ; Task complexity ; Vision ; Visual tasks</subject><ispartof>arXiv.org, 2024-11</ispartof><rights>2024. This work is published under http://arxiv.org/licenses/nonexclusive-distrib/1.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</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>780,784</link.rule.ids></links><search><creatorcontrib>Wang, Taowen</creatorcontrib><creatorcontrib>Liu, Dongfang</creatorcontrib><creatorcontrib>Liang, James Chenhao</creatorcontrib><creatorcontrib>Yang, Wenhao</creatorcontrib><creatorcontrib>Wang, Qifan</creatorcontrib><creatorcontrib>Cheng, Han</creatorcontrib><creatorcontrib>Luo, Jiebo</creatorcontrib><creatorcontrib>Tang, Ruixiang</creatorcontrib><title>Exploring the Adversarial Vulnerabilities of Vision-Language-Action Models in Robotics</title><title>arXiv.org</title><description>Recently in robotics, Vision-Language-Action (VLA) models have emerged as a transformative approach, enabling robots to execute complex tasks by integrating visual and linguistic inputs within an end-to-end learning framework. While VLA models offer significant capabilities, they also introduce new attack surfaces, making them vulnerable to adversarial attacks. With these vulnerabilities largely unexplored, this paper systematically quantifies the robustness of VLA-based robotic systems. Recognizing the unique demands of robotic execution, our attack objectives target the inherent spatial and functional characteristics of robotic systems. In particular, we introduce an untargeted position-aware attack objective that leverages spatial foundations to destabilize robotic actions, and a targeted attack objective that manipulates the robotic trajectory. Additionally, we design an adversarial patch generation approach that places a small, colorful patch within the camera's view, effectively executing the attack in both digital and physical environments. Our evaluation reveals a marked degradation in task success rates, with up to a 100\% reduction across a suite of simulated robotic tasks, highlighting critical security gaps in current VLA architectures. By unveiling these vulnerabilities and proposing actionable evaluation metrics, this work advances both the understanding and enhancement of safety for VLA-based robotic systems, underscoring the necessity for developing robust defense strategies prior to physical-world deployments.</description><subject>Robotics</subject><subject>Task complexity</subject><subject>Vision</subject><subject>Visual tasks</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>ABUWG</sourceid><sourceid>AFKRA</sourceid><sourceid>AZQEC</sourceid><sourceid>BENPR</sourceid><sourceid>CCPQU</sourceid><sourceid>DWQXO</sourceid><recordid>eNqNikELgjAYQEcQJOV_GHQeuC3LjhJGh7pEeJWp0z4Zm-2b0c_PQz-g0-Px3oJEQkrOsp0QKxIjDkmSiP1BpKmMSFl8RuM82J6Gp6Z5-9YelQdlaDkZq72qwUAAjdR1tAQEZ9lV2X5SvWZ5E2anN9dqgxQsvbvaBWhwQ5adMqjjH9dkey4epwsbvXtNGkM1uMnbOVWSS35MOZeZ_O_6AjlyQWM</recordid><startdate>20241122</startdate><enddate>20241122</enddate><creator>Wang, Taowen</creator><creator>Liu, Dongfang</creator><creator>Liang, James Chenhao</creator><creator>Yang, Wenhao</creator><creator>Wang, Qifan</creator><creator>Cheng, Han</creator><creator>Luo, Jiebo</creator><creator>Tang, Ruixiang</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope></search><sort><creationdate>20241122</creationdate><title>Exploring the Adversarial Vulnerabilities of Vision-Language-Action Models in Robotics</title><author>Wang, Taowen ; Liu, Dongfang ; Liang, James Chenhao ; Yang, Wenhao ; Wang, Qifan ; Cheng, Han ; Luo, Jiebo ; Tang, Ruixiang</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_31319511383</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Robotics</topic><topic>Task complexity</topic><topic>Vision</topic><topic>Visual tasks</topic><toplevel>online_resources</toplevel><creatorcontrib>Wang, Taowen</creatorcontrib><creatorcontrib>Liu, Dongfang</creatorcontrib><creatorcontrib>Liang, James Chenhao</creatorcontrib><creatorcontrib>Yang, Wenhao</creatorcontrib><creatorcontrib>Wang, Qifan</creatorcontrib><creatorcontrib>Cheng, Han</creatorcontrib><creatorcontrib>Luo, Jiebo</creatorcontrib><creatorcontrib>Tang, Ruixiang</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni Edition)</collection><collection>ProQuest Central UK/Ireland</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central Korea</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Access via ProQuest (Open Access)</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Wang, Taowen</au><au>Liu, Dongfang</au><au>Liang, James Chenhao</au><au>Yang, Wenhao</au><au>Wang, Qifan</au><au>Cheng, Han</au><au>Luo, Jiebo</au><au>Tang, Ruixiang</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Exploring the Adversarial Vulnerabilities of Vision-Language-Action Models in Robotics</atitle><jtitle>arXiv.org</jtitle><date>2024-11-22</date><risdate>2024</risdate><eissn>2331-8422</eissn><abstract>Recently in robotics, Vision-Language-Action (VLA) models have emerged as a transformative approach, enabling robots to execute complex tasks by integrating visual and linguistic inputs within an end-to-end learning framework. While VLA models offer significant capabilities, they also introduce new attack surfaces, making them vulnerable to adversarial attacks. With these vulnerabilities largely unexplored, this paper systematically quantifies the robustness of VLA-based robotic systems. Recognizing the unique demands of robotic execution, our attack objectives target the inherent spatial and functional characteristics of robotic systems. In particular, we introduce an untargeted position-aware attack objective that leverages spatial foundations to destabilize robotic actions, and a targeted attack objective that manipulates the robotic trajectory. Additionally, we design an adversarial patch generation approach that places a small, colorful patch within the camera's view, effectively executing the attack in both digital and physical environments. Our evaluation reveals a marked degradation in task success rates, with up to a 100\% reduction across a suite of simulated robotic tasks, highlighting critical security gaps in current VLA architectures. By unveiling these vulnerabilities and proposing actionable evaluation metrics, this work advances both the understanding and enhancement of safety for VLA-based robotic systems, underscoring the necessity for developing robust defense strategies prior to physical-world deployments.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2024-11
issn 2331-8422
language eng
recordid cdi_proquest_journals_3131951138
source Free E- Journals
subjects Robotics
Task complexity
Vision
Visual tasks
title Exploring the Adversarial Vulnerabilities of Vision-Language-Action Models in Robotics
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-23T13%3A45%3A37IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=Exploring%20the%20Adversarial%20Vulnerabilities%20of%20Vision-Language-Action%20Models%20in%20Robotics&rft.jtitle=arXiv.org&rft.au=Wang,%20Taowen&rft.date=2024-11-22&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E3131951138%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=3131951138&rft_id=info:pmid/&rfr_iscdi=true