Google DeepMind unveils Gemini AI that can control humanoid robots

Google DeepMind has just unveiled a version of the Gemini AI model aimed at controlling humanoid robots for the very first time.

Named Gemini Robotics 2, the system is a blend of vision language model (VLM) with two vision language action (VLA) models. The combination allows robots to communicate with humans by understanding the surroundings and perform tasks autonomously.

What makes the system more distinguished is “intelligent whole-body control” that helps humanoids to carry out various tasks. According to Google DeepMind, Gemini Robotics 2 is a milestone towards achieving “physical AGI.”

“Most robots are pre-programmed or teleoperated for narrow, repetitive task sequences. They lack the ability to truly learn for themselves or adapt to unpredictable environments,” Carolina Parada, head of robotics at Google DeepMind, wrote in a blog post.

“For the first time, our model can now control entire humanoid robots, translating intent into intelligent whole-body control.”

"By building this core intelligence, our goal is to enable AI in the physical world that can work alongside humans to solve complex challenges,” Parada added.

Google DeepMind CEO Demis Hassabis also laid out a vision based on developing an AI system that would serve as an Android operating system for robots.

Standardizing the robotic "brain" allows companies such as Boston Dynamics to concentrate on hardware engineering by leveraging unified AI architectures for perception, reasoning, and control.