DeepMind Introduces Advanced AI Models for Next-Gen Robotics

Google DeepMind unveils two new AI models, Gemini Robotics and Gemini Robotics-ER, aiming to revolutionize robotics with enhanced adaptability, interactivity, and dexterity.

Google DeepMind just dropped a bombshell: they’ve cooked up two groundbreaking AI models, Gemini Robotics and Gemini Robotics-ER, building on the smarts of Gemini 2.0. These aren’t your average robots—they’re about to turn the whole game on its head by nailing three big things: being general, interactive, and downright dexterous.

Generality? That’s all about a robot rolling with the punches, handling stuff it’s never seen before like a pro. Interactivity means it gets you—responding to your voice or a sudden change in its surroundings without missing a beat. And dexterity? Well, that’s the robot showing off its fine motor skills, doing things we used to think were too delicate for machines (like folding origami without turning it into a paper ball).

Take the ALOHA 2 robot, for example. Thanks to Gemini Robotics, it’s now folding origami and sealing Ziploc bags like it’s been doing it its whole life—all while chatting away in natural language. It’s not just cool; it’s a giant leap toward robots that can do, well, pretty much anything.

And there’s more. Google’s teaming up with Apptronik (the brains behind the Apollo bipedal robot) to dream up the next wave of humanoid robots. Plus, with Gemini Robotics-ER hitting the scene, roboticists everywhere will get to play with Gemini’s top-tier reasoning skills. And guess who’s first in line? Boston Dynamics, because of course they are.