Google DeepMind AI Controls Robot's Entire Body

Google DeepMind's new AI model, Gemini Robotics 2, can control entire humanoid robots.

Google DeepMind has announced that the latest version of its Gemini Robotics AI model can now control entire humanoid robots. This new iteration, Gemini Robotics 2, supports "whole-body motions" from feet to fingertips, a significant expansion from the previous model's focus on upper body control. The model enables robots to perform a wider range of actions, including walking, crouching, stretching, and manipulating objects.

Gemini Robotics 2 also brings improvements in dexterity, allowing for control of complex, five-fingered hands for tasks like sealing a Ziploc or tying a trash bag. Google DeepMind is also updating Gemini Robotics ER (embodied reasoning), a vision-language model that helps robots analyze surroundings and perform multi-step tasks. Gemini Robotics ER 2 is better at completing extended tasks and understands task beginnings and endings, and can coordinate multiple robot types.

This update is considered an important step toward enabling robots to complete more complex, real-world tasks requiring whole-body coordination. Google DeepMind notes that Gemini Robotics ER 2 is its "safest robotics model to date" due to improved human detection and safety stops. Additionally, the Gemini Robotics On-Device Model, which runs locally, can now adapt faster to new robot embodiments.

Find the right AI tool for your business

Chat with Insta and get matched to the right tool in seconds.