Google DeepMind unveils Gemini Robotics 2 for whole-body robot control

Gemini Robotics 2 arrives as part of a broader wave of robotics demonstrations and research highlighted in this week's IEEE Spectrum roundup. The system is framed as an "intelligence layer" — a software abstraction that sits atop robot hardware to coordinate movement and enable adaptive behavior. The article positions this as a significant enough advance to warrant calling it a "major advance," suggesting that moving from scripted, single-task robots to systems capable of intelligent whole-body control and multi-robot coordination represents a meaningful threshold in robotics capability.

The broader robotics landscape shown in the article—spanning humanoid motion transfer, dexterous manipulation with soft materials, planetary exploration robots, and household navigation—suggests that the field is pursuing robotic capability across multiple form factors and domains. Gemini Robotics 2 appears to be positioned as a general AI component that could be deployed across many of these applications, rather than a solution tailored to one task or body shape.