GR-2: A Generative Video-Language-Action Model with Web-Scale Knowledge for Robot Manipulation

We present GR-2, a state-of-the-art generalist robot agent for versatile and generalizable robot manipulation. GR-2 is first pre-trained on a vast number of Internet videos to capture the dynamics of the world. This large-scale pre-training, involving 38 million video clips and over 50 billion token...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2024-10
Hauptverfasser: Chi-Lam Cheang, Chen, Guangzeng, Jing, Ya, Kong, Tao, Li, Hang, Li, Yifeng, Liu, Yuxiao, Wu, Hongtao, Xu, Jiafeng, Yang, Yichu, Zhang, Hanbo, Zhu, Minzhao
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 Chi-Lam Cheang
Chen, Guangzeng
Jing, Ya
Kong, Tao
Li, Hang
Li, Yifeng
Liu, Yuxiao
Wu, Hongtao
Xu, Jiafeng
Yang, Yichu
Zhang, Hanbo
Zhu, Minzhao
description We present GR-2, a state-of-the-art generalist robot agent for versatile and generalizable robot manipulation. GR-2 is first pre-trained on a vast number of Internet videos to capture the dynamics of the world. This large-scale pre-training, involving 38 million video clips and over 50 billion tokens, equips GR-2 with the ability to generalize across a wide range of robotic tasks and environments during subsequent policy learning. Following this, GR-2 is fine-tuned for both video generation and action prediction using robot trajectories. It exhibits impressive multi-task learning capabilities, achieving an average success rate of 97.7% across more than 100 tasks. Moreover, GR-2 demonstrates exceptional generalization to new, previously unseen scenarios, including novel backgrounds, environments, objects, and tasks. Notably, GR-2 scales effectively with model size, underscoring its potential for continued growth and application. Project page: \url{https://gr2-manipulation.github.io}.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_3115224336</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>3115224336</sourcerecordid><originalsourceid>FETCH-proquest_journals_31152243363</originalsourceid><addsrcrecordid>eNqNi8sKgkAUQIcgKKp_uNB6QGe0ol1EDyg3FrVLRr3ayDDXdKzfr6APaHUW55weGwopfb4IhBiwSdtWnueJ2VyEoRyy2y7mYgkr2KHFRjn9RLjoHIkflS07VSJfZU6ThYhyNPDS7g5XTPkpUwbhYOllMC8RCmogppQcRMrqujPqe41Zv1CmxcmPIzbdbs7rPa8benTYuqSirrEflUjfD4UIpJzJ_6o3M8RCbQ</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>3115224336</pqid></control><display><type>article</type><title>GR-2: A Generative Video-Language-Action Model with Web-Scale Knowledge for Robot Manipulation</title><source>Free E- Journals</source><creator>Chi-Lam Cheang ; Chen, Guangzeng ; Jing, Ya ; Kong, Tao ; Li, Hang ; Li, Yifeng ; Liu, Yuxiao ; Wu, Hongtao ; Xu, Jiafeng ; Yang, Yichu ; Zhang, Hanbo ; Zhu, Minzhao</creator><creatorcontrib>Chi-Lam Cheang ; Chen, Guangzeng ; Jing, Ya ; Kong, Tao ; Li, Hang ; Li, Yifeng ; Liu, Yuxiao ; Wu, Hongtao ; Xu, Jiafeng ; Yang, Yichu ; Zhang, Hanbo ; Zhu, Minzhao</creatorcontrib><description>We present GR-2, a state-of-the-art generalist robot agent for versatile and generalizable robot manipulation. GR-2 is first pre-trained on a vast number of Internet videos to capture the dynamics of the world. This large-scale pre-training, involving 38 million video clips and over 50 billion tokens, equips GR-2 with the ability to generalize across a wide range of robotic tasks and environments during subsequent policy learning. Following this, GR-2 is fine-tuned for both video generation and action prediction using robot trajectories. It exhibits impressive multi-task learning capabilities, achieving an average success rate of 97.7% across more than 100 tasks. Moreover, GR-2 demonstrates exceptional generalization to new, previously unseen scenarios, including novel backgrounds, environments, objects, and tasks. Notably, GR-2 scales effectively with model size, underscoring its potential for continued growth and application. Project page: \url{https://gr2-manipulation.github.io}.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Robot learning ; Robots</subject><ispartof>arXiv.org, 2024-10</ispartof><rights>2024. This work is published under http://creativecommons.org/licenses/by/4.0/ (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>Chi-Lam Cheang</creatorcontrib><creatorcontrib>Chen, Guangzeng</creatorcontrib><creatorcontrib>Jing, Ya</creatorcontrib><creatorcontrib>Kong, Tao</creatorcontrib><creatorcontrib>Li, Hang</creatorcontrib><creatorcontrib>Li, Yifeng</creatorcontrib><creatorcontrib>Liu, Yuxiao</creatorcontrib><creatorcontrib>Wu, Hongtao</creatorcontrib><creatorcontrib>Xu, Jiafeng</creatorcontrib><creatorcontrib>Yang, Yichu</creatorcontrib><creatorcontrib>Zhang, Hanbo</creatorcontrib><creatorcontrib>Zhu, Minzhao</creatorcontrib><title>GR-2: A Generative Video-Language-Action Model with Web-Scale Knowledge for Robot Manipulation</title><title>arXiv.org</title><description>We present GR-2, a state-of-the-art generalist robot agent for versatile and generalizable robot manipulation. GR-2 is first pre-trained on a vast number of Internet videos to capture the dynamics of the world. This large-scale pre-training, involving 38 million video clips and over 50 billion tokens, equips GR-2 with the ability to generalize across a wide range of robotic tasks and environments during subsequent policy learning. Following this, GR-2 is fine-tuned for both video generation and action prediction using robot trajectories. It exhibits impressive multi-task learning capabilities, achieving an average success rate of 97.7% across more than 100 tasks. Moreover, GR-2 demonstrates exceptional generalization to new, previously unseen scenarios, including novel backgrounds, environments, objects, and tasks. Notably, GR-2 scales effectively with model size, underscoring its potential for continued growth and application. Project page: \url{https://gr2-manipulation.github.io}.</description><subject>Robot learning</subject><subject>Robots</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>eNqNi8sKgkAUQIcgKKp_uNB6QGe0ol1EDyg3FrVLRr3ayDDXdKzfr6APaHUW55weGwopfb4IhBiwSdtWnueJ2VyEoRyy2y7mYgkr2KHFRjn9RLjoHIkflS07VSJfZU6ThYhyNPDS7g5XTPkpUwbhYOllMC8RCmogppQcRMrqujPqe41Zv1CmxcmPIzbdbs7rPa8benTYuqSirrEflUjfD4UIpJzJ_6o3M8RCbQ</recordid><startdate>20241008</startdate><enddate>20241008</enddate><creator>Chi-Lam Cheang</creator><creator>Chen, Guangzeng</creator><creator>Jing, Ya</creator><creator>Kong, Tao</creator><creator>Li, Hang</creator><creator>Li, Yifeng</creator><creator>Liu, Yuxiao</creator><creator>Wu, Hongtao</creator><creator>Xu, Jiafeng</creator><creator>Yang, Yichu</creator><creator>Zhang, Hanbo</creator><creator>Zhu, Minzhao</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>20241008</creationdate><title>GR-2: A Generative Video-Language-Action Model with Web-Scale Knowledge for Robot Manipulation</title><author>Chi-Lam Cheang ; Chen, Guangzeng ; Jing, Ya ; Kong, Tao ; Li, Hang ; Li, Yifeng ; Liu, Yuxiao ; Wu, Hongtao ; Xu, Jiafeng ; Yang, Yichu ; Zhang, Hanbo ; Zhu, Minzhao</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_31152243363</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Robot learning</topic><topic>Robots</topic><toplevel>online_resources</toplevel><creatorcontrib>Chi-Lam Cheang</creatorcontrib><creatorcontrib>Chen, Guangzeng</creatorcontrib><creatorcontrib>Jing, Ya</creatorcontrib><creatorcontrib>Kong, Tao</creatorcontrib><creatorcontrib>Li, Hang</creatorcontrib><creatorcontrib>Li, Yifeng</creatorcontrib><creatorcontrib>Liu, Yuxiao</creatorcontrib><creatorcontrib>Wu, Hongtao</creatorcontrib><creatorcontrib>Xu, Jiafeng</creatorcontrib><creatorcontrib>Yang, Yichu</creatorcontrib><creatorcontrib>Zhang, Hanbo</creatorcontrib><creatorcontrib>Zhu, Minzhao</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>Publicly Available Content Database</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>Chi-Lam Cheang</au><au>Chen, Guangzeng</au><au>Jing, Ya</au><au>Kong, Tao</au><au>Li, Hang</au><au>Li, Yifeng</au><au>Liu, Yuxiao</au><au>Wu, Hongtao</au><au>Xu, Jiafeng</au><au>Yang, Yichu</au><au>Zhang, Hanbo</au><au>Zhu, Minzhao</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>GR-2: A Generative Video-Language-Action Model with Web-Scale Knowledge for Robot Manipulation</atitle><jtitle>arXiv.org</jtitle><date>2024-10-08</date><risdate>2024</risdate><eissn>2331-8422</eissn><abstract>We present GR-2, a state-of-the-art generalist robot agent for versatile and generalizable robot manipulation. GR-2 is first pre-trained on a vast number of Internet videos to capture the dynamics of the world. This large-scale pre-training, involving 38 million video clips and over 50 billion tokens, equips GR-2 with the ability to generalize across a wide range of robotic tasks and environments during subsequent policy learning. Following this, GR-2 is fine-tuned for both video generation and action prediction using robot trajectories. It exhibits impressive multi-task learning capabilities, achieving an average success rate of 97.7% across more than 100 tasks. Moreover, GR-2 demonstrates exceptional generalization to new, previously unseen scenarios, including novel backgrounds, environments, objects, and tasks. Notably, GR-2 scales effectively with model size, underscoring its potential for continued growth and application. Project page: \url{https://gr2-manipulation.github.io}.</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, 2024-10
issn 2331-8422
language eng
recordid cdi_proquest_journals_3115224336
source Free E- Journals
subjects Robot learning
Robots
title GR-2: A Generative Video-Language-Action Model with Web-Scale Knowledge for Robot Manipulation
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-09T21%3A54%3A26IST&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=GR-2:%20A%20Generative%20Video-Language-Action%20Model%20with%20Web-Scale%20Knowledge%20for%20Robot%20Manipulation&rft.jtitle=arXiv.org&rft.au=Chi-Lam%20Cheang&rft.date=2024-10-08&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E3115224336%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=3115224336&rft_id=info:pmid/&rfr_iscdi=true