{"id":30588,"date":"2025-03-13T10:37:24","date_gmt":"2025-03-13T02:37:24","guid":{"rendered":"https:\/\/www.1ai.net\/?p=30588"},"modified":"2025-03-13T10:37:24","modified_gmt":"2025-03-13T02:37:24","slug":"%e8%b0%b7%e6%ad%8c-deepmind-%e6%8e%a8%e5%87%ba%e6%96%b0-ai-%e6%a8%a1%e5%9e%8b%ef%bc%8c%e6%9c%ba%e5%99%a8%e4%ba%ba%e6%9c%aa%e7%bb%8f%e8%ae%ad%e7%bb%83%e4%b9%9f%e8%83%bd%e6%89%a7%e8%a1%8c%e7%8e%b0","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/30588.html","title":{"rendered":"Google DeepMind Introduces New AI Models That Let Robots Perform Real-World Tasks Without Training"},"content":{"rendered":"

On the evening of March 12, Beijing time.Google<\/a> DeepMind<\/a> Launch of two new models AI Models<\/a>that is designed to help robots accomplish moreMultiple real-world missions.<\/strong><\/p>\n

\"cd6a42d4j00st1j7t00e1d000u000k3p\"<\/p>\n

One of them, called Gemini Robotics, is avisual language action model<\/strong>The ability to make robotsUnderstanding new situations without specialized training<\/strong>.<\/p>\n

Gemini Robotics is based on the latest version of Google's flagship AI model, Gemini 2.0, which builds on Gemini's multimodal world understanding capabilities by adding new modalities for physical action, according to Carolina Parada, senior director of robotics at Google DeepMind. Gemini Robotics builds on Gemini's multimodal world-understanding capabilities and applies them to the real world by adding new modalities for physical action.<\/p>\n

The model makes progress in the three core areas that Google DeepMind believes are necessary to build efficient robots: versatility, interaction, and flexibility. In addition to being able to cope with new contexts, Gemini Robotics performs better at interacting with humans and the environment, and is able to perform more precise physical operations, such asFolding paper or opening bottle caps<\/strong>.<\/p>\n

The other is the Gemini Robotics-ER (Embodied Reasoning) model, which the company describes as an advanced visual language model capable of \"Understanding a complex and dynamic world<\/strong>\u201d.<\/p>\n

Parada further explains that when you're filling a bento box.Where to place items on the table and how to do it<\/strong>The Gemini Robotics-ER is designed for this type of reasoning task, and robotics experts can use the model to interface with existing low-level control systems, opening up new capabilities driven by the Gemini Robotics-ER.<\/p>\n

Vikas Sindhwani, a researcher at Google DeepMind, said the company is developing a \"layered security policy\" and said the Gemini Robotics-ER model has been trained to assess whether an action is safe or not in a given situation. The company has also released new benchmarks and frameworks to advance security research in AI. According to 1AI, last year, Google DeepMind Introduced the \"Robot Constitution\"<\/strong>, which is a code of conduct for robots inspired by Isaac Asimov.<\/p>\n

According to The Verge, Google DeepMind has partnered with Apptronik to \"build the next generation of humanoid robots\". In addition, Google has opened up the Gemini Robotics-ER model to \"trusted testers\" including Agile Robots, Agility Robotics, Boston Dynamics and Enchanted Tools, Parada said: \"We're focused on building intelligence that understands and acts in the physical world, and we're looking forward to applying this technology to multiple domains and manifestations.\"<\/p>","protected":false},"excerpt":{"rendered":"

On the evening of March 12, Beijing time, Google DeepMind launched two new AI models designed to help robots perform more tasks in the real world. One of them, Gemini Robotics, is a visual linguistic action model that enables robots to understand new situations without specialized training. Gemini Robotics is based on Google's latest version of the AI flagship model - Gemini 2.0. Google DeepMind, the senior director of robotics, Carolina Parada, has said that Gemini Robotics, relying on Gemini's multi-modular world understanding capabilities, has joined the new paradigm of physical action<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[146],"tags":[167,593,281],"collection":[],"class_list":["post-30588","post","type-post","status-publish","format-standard","hentry","category-news","tag-ai","tag-deepmind","tag-281"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/30588","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=30588"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/30588\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=30588"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=30588"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=30588"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=30588"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}