New AI for robots — Google unveils Gemini Robotics
Share
Google DeepMind development team is constantly developing the field of AI by updating Gemini, Imagen, Veo, Gemma, and AlphaFold models. This week, the company also entered the robotics market by unveiling Gemini Robotics.
Neowin writes about it.
Everything we know about Google Gemini Robotics model
Gemini Robotics is an advanced Voice-Language-Action model based on Gemini 2.0 AI, which allows controlling robots by performing physical actions. Google claims that the model can understand situations that it has not previously encountered during training.
The new model performs twice as well on a comprehensive generalisation test compared to other state-of-the-art models on the market. Given that Gemini Robotics has the ability to understand natural language in different languages, robots can understand human teams much better.
Google says the new model can perform complex and multi-stage tasks that require precise manipulation. For example, it can fold origami or put a snack in a ziplock bag.
Google has also unveiled the Gemini Robotics-ER model, which focuses on spatial thinking and allows robotics experts to connect it to existing low-level controllers. With this model, the robot can be controlled right out of the box.
Google has partnered with Apptronik to create humanoid robots.
As a reminder, Gemini AI is available to all users of modern devices. The chatbot has powerful features that can improve life.
We also wrote that iPhone users were able to communicate with Google Gemini from the lock screen. Google has added the Gemini Live widget with the new update of the chatbot for iOS.
Read Novyny.LIVE!
