{"id":14470,"date":"2025-09-26T18:00:00","date_gmt":"2025-09-26T16:00:00","guid":{"rendered":"https:\/\/haimagazine.com\/uncategorized\/an-update-that-teaches-robots-to-plan-and-use-the-internet\/"},"modified":"2025-10-02T14:08:24","modified_gmt":"2025-10-02T12:08:24","slug":"an-update-that-teaches-robots-to-plan-and-use-the-internet","status":"publish","type":"post","link":"https:\/\/haimagazine.com\/en\/uncategorized\/an-update-that-teaches-robots-to-plan-and-use-the-internet\/","title":{"rendered":"\ud83d\udd12 An update that teaches robots to plan and use the internet"},"content":{"rendered":"<p>Google DeepMind has unveiled the new generation of its AI models for robotics: <strong>Gemini Robotics 1.5<\/strong> and <strong>Gemini Robotics-ER 1.5<\/strong>. It&#8217;s a significant expansion of the solutions from March 2024, <strong>giving robots the ability to perform multi-stage planning<\/strong> and <strong>use the internet to supplement data<\/strong>. Thanks to the new models, machines can plan actions several steps ahead, allowing them to tackle complex tasks in the physical world.<\/p><h4 class=\"wp-block-heading\"><strong>How does it work? Two-stage architecture<\/strong><\/h4><p>The heart of the system is the collaboration of two specialized models. The first one, Gemini Robotics-ER 1.5, serves as a component for analysis and planning. It processes data about the environment and the goal, uses digital tools to find necessary information (like local regulations), and then generates an action plan in natural language. These instructions are sent to the second model, Gemini Robotics 1.5, which is a <strong>VLA<\/strong>-type (vision-language-action) execution system that translates the received plan into specific, physical robot operations.<strong> This architecture signifies a fundamental change: moving from executing simple, single commands to autonomously carrying out multi-step tasks.<\/strong><\/p><h4 class=\"wp-block-heading\"><strong>New skills: from sorting to recycling based on local regulations<\/strong><\/h4><p>The new robot capabilities were illustrated with practical examples. They can sort laundry, pack a suitcase after checking the weather forecast at the destination, and even help recycle according to local guidelines found online. These tasks require not only object recognition but, most importantly, access to external data and a sequential planning of actions.<\/p><h4 class=\"wp-block-heading\"><strong>Skill transfer between robots<\/strong><\/h4><p>One of the biggest breakthroughs is <strong>the ability to transfer skills between robots of completely different structures<\/strong>. The Google DeepMind team showed that a model trained on the two-armed ALOHA2 robot can successfully be used to control the Frank robot or the Apollo humanoid. Such versatility allows us to overcome one of the biggest challenges in robotics \u2014 the need to spend a fortune training models from scratch for each new hardware platform.<\/p><h4 class=\"wp-block-heading\"><strong>Availability for developers<\/strong><\/h4><p>Google is rolling out new tools for developers in a thoughtful way. Gemini Robotics-ER 1.5 \u2014 the model responsible for analysis and planning \u2014 is publicly available through the Gemini API in Google AI Studio. Meanwhile, Gemini Robotics 1.5 \u2014 the executive model \u2014 is currently only available to selected partners, which allows for more controlled deployments in the physical layer.<\/p><h4 class=\"wp-block-heading\"><strong>Challenges and perspectives<\/strong><\/h4><p>Despite visible progress, experts point out there are still challenges to overcome. Issues like manual dexterity, safety in human interaction, and reliability in varying conditions remain key areas for further research. Meeting the standards required in industrial applications will take many more tests.<\/p><p>The new Google DeepMind models are setting an important direction for the development of robotics. For engineers, they offer the chance to create systems that dynamically plan tasks, combining perception with internet data. They also reduce implementation costs due to the portability of software across different platforms. The launch of Gemini Robotics 1.5 is a step towards machines that not only execute commands but also interpret context and plan their actions independently.<\/p>","protected":false},"excerpt":{"rendered":"<p>Google DeepMind has unveiled models that combine reasoning in the physical world with online search and skill transfer between robots. The goal is to achieve multi-step tasks and fewer scripts.<\/p>\n","protected":false},"author":230,"featured_media":14249,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"rank_math_lock_modified_date":false,"footnotes":""},"categories":[813,1],"tags":[707,749,747],"popular":[],"difficulty-level":[36],"ppma_author":[884],"class_list":["post-14470","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-news-2","category-uncategorized","tag-ai-5","tag-deepmind-2","tag-google-2","difficulty-level-easy"],"acf":[],"authors":[{"term_id":884,"user_id":230,"is_guest":0,"slug":"karolina-ceron","display_name":"Karolina Cero\u0144","avatar_url":"https:\/\/haimagazine.com\/wp-content\/uploads\/2025\/07\/PXL_20250419_110132091.MP4-scaled.jpg","first_name":"Karolina","last_name":"Cero\u0144","user_url":"","job_title":"","description":"Wsp\u00f3\u0142tw\u00f3rczyni newslettera AI Flash, studentka psychologii i pasjonatka sztucznej inteligencji. Interesuj\u0119 si\u0119 wp\u0142ywem nowych technologii na cz\u0142owieka, a w wolnych chwilach eksperymentuj\u0119 z generatywn\u0105 grafik\u0105 w Midjourney."}],"_links":{"self":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/14470","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/users\/230"}],"replies":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/comments?post=14470"}],"version-history":[{"count":1,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/14470\/revisions"}],"predecessor-version":[{"id":14471,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/14470\/revisions\/14471"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media\/14249"}],"wp:attachment":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media?parent=14470"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/categories?post=14470"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/tags?post=14470"},{"taxonomy":"popular","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/popular?post=14470"},{"taxonomy":"difficulty-level","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/difficulty-level?post=14470"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/ppma_author?post=14470"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}