The Sequence AI of the Week #908: You Need to Learn About Gemini Robotics
Google put a walking humanoid inside a single policy, published the reasoning half as an API, and kept the motor half behind a partner gate. The numbers explain why.
Gemini Robotics 2.0 features a significant advancement where walking and grasping are controlled by the same policy, conditioned on a single language instruction. This approach integrates locomotion into the action space, a departure from older models that treated them as separate subsystems. This unified control is the key innovation of Gemini Robotics 2.
- Gemini Robotics 2.0’s demo is intentionally simple to highlight its capabilities.
- Previous Gemini Robotics models were limited to tabletop tasks with fixed bases or external leg controllers.
- The innovation lies in a single policy controlling both walking and grasping, unified by language instructions.
- Locomotion is now part of the predicted action space, not a separate subsystem.
- This integration represents a significant advancement in AI robotics.
https://bender.layer3.press/articles/4c86aae5-e3f1-4827-95fe-0e67560a396c
Write a comment