{"id":25592,"date":"2025-07-25T17:39:18","date_gmt":"2025-07-25T14:39:18","guid":{"rendered":"https:\/\/forklog.com\/en\/bytedance-unveils-ai-brain-for-robots\/"},"modified":"2025-07-25T17:39:18","modified_gmt":"2025-07-25T14:39:18","slug":"bytedance-unveils-ai-brain-for-robots","status":"publish","type":"post","link":"https:\/\/forklog.com\/en\/bytedance-unveils-ai-brain-for-robots\/","title":{"rendered":"ByteDance Unveils AI &#8220;Brain&#8221; for Robots"},"content":{"rendered":"<p>ByteDance, the company behind TikTok, has <a href=\"https:\/\/seed.bytedance.com\/en\/GR3\">introduced<\/a> a system that acts as a &#8220;brain&#8221; for robots, enabling them to perform household tasks such as hanging clothes or clearing tables.<\/p>\n<p>GR-3 is a large vision-language-action model that allows robots to follow natural language commands and perform general tasks with unfamiliar objects. They can operate in new environments or with abstract concepts related to size and spatial relationships.<\/p>\n<p>A video published on the website demonstrates how ByteMini, a lab-based two-armed robot, can insert a hanger into a shirt and place it on a rack.<\/p>\n<blockquote class=\"twitter-tweet\">\n<p lang=\"en\" dir=\"ltr\">\ud83d\ude80\ud83d\ude80\ud83d\ude80 Ever wondered what it takes for robots to handle real-world household tasks? long-horizon execution, deformable object dexterity, and unseen object generalization \u2014 meet GR-3, ByteDance Seed\u2019s new Vision-Language-Action (VLA) model!<\/p>\n<p>GR-3 is a generalizable\u2026 <a href=\"https:\/\/t.co\/zECRjaXC0J\">pic.twitter.com\/zECRjaXC0J<\/a><\/p>\n<p>\u2014 Xiao Ma (@yusufma555) <a href=\"https:\/\/twitter.com\/yusufma555\/status\/1947504955272925213?ref_src=twsrc%5Etfw\">July 22, 2025<\/a><\/p><\/blockquote>\n<p> <script async src=\"https:\/\/platform.twitter.com\/widgets.js\" charset=\"utf-8\"><\/script><\/p>\n<p>In a separate technical report, the team noted that the robot can handle short-sleeved clothing, even though &#8220;all items in the training data had long sleeves.&#8221;<\/p>\n<p>Thanks to GR-3, the robot can execute commands to select a specific item from several and place it in a designated location.<\/p>\n<p>The system can recognize an object not only by name but also by size (e.g., &#8220;large plate&#8221;) or spatial attribute (e.g., &#8220;on the left&#8221;). It can fully execute the task of &#8220;clearing the dining table&#8221; with a single command.<\/p>\n<p>To train the model, ByteDance employed a multi-component approach, including:<\/p>\n<ul class=\"wp-block-list\">\n<li>joint training on large datasets in the &#8220;image-text&#8221; format; <\/li>\n<li>fine-tuning on human action trajectory data collected via VR devices;<\/li>\n<li>imitation learning on android movement data.<\/li>\n<\/ul>\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>&#8220;We hope that GR-3 will be a step towards creating universal robots capable of assisting people in everyday life,&#8221; the team stated.<\/p>\n<\/blockquote>\n<p>Back in January, the startup Perplexity AI <a href=\"https:\/\/forklog.com\/en\/news\/reports-perplexity-ai-proposes-merger-with-tiktok-u-s\">announced its intention<\/a> to acquire the American TikTok. The firm sent ByteDance a proposal to merge Perplexity, TikTok U.S., and new equity partners into a single legal entity.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>ByteDance, the company behind TikTok, has introduced a system that acts as a &#8220;brain&#8221; for robots, enabling them to perform household tasks such as hanging clothes or clearing tables. GR-3 is a large vision-language-action model that allows robots to follow natural language commands and perform general tasks with unfamiliar objects. They can operate in new [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":25591,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"select":"","news_style_id":"","cryptorium_level":"","_short_excerpt_text":"","creation_source":"","_metatest_mainpost_news_update":false,"footnotes":""},"categories":[3],"tags":[438,1589],"class_list":["post-25592","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news-and-analysis","tag-artificial-intelligence","tag-tiktok"],"aioseo_notices":[],"amp_enabled":true,"views":"65","promo_type":"","layout_type":"","short_excerpt":"","is_update":"","_links":{"self":[{"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/posts\/25592","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/comments?post=25592"}],"version-history":[{"count":0,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/posts\/25592\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/media\/25591"}],"wp:attachment":[{"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/media?parent=25592"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/categories?post=25592"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/forklog.com\/en\/wp-json\/wp\/v2\/tags?post=25592"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}