Google DeepMind's Gemini Robotics ER 2 plans and monitors robot work while lower-level controllers carry out movement. Developers expose movement or navigation interfaces as tools and stream observations into the model. This lets it reason about the next step while the robot is acting. Announcement
Continuous video also gives the system information about whether an action is progressing or has failed. The release describes using that feedback to adjust actions or retry a step without restarting the entire workflow. The contribution is coordination between observation, planning and existing robot controls, with physical execution still handled by those controls.
Terms in this piece · Glossary
reasoning model — A model trained to think — generating extended internal reasoning before answering — trading time and tokens for accuracy on hard problems.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
