Gemini Robotics: Whole-Body Control for Real-World Robot Tasks

This video demonstrates how the Gemini Robotics embodied reasoning model enables a humanoid robot to understand natural language instructions and visual scenes, then coordinate its entire body to perform complex tasks in a real-world environment.

- Robot helping pack sports equipment for children's activities
- Integration of embodied reasoning with a vision-language-action model for whole-body control
- Navigating cluttered environments, retrieving objects, and placing them correctly
- Recognizing failures and retrying, plus a reactivity stress test with an unexpected object

This video is suitable for those interested in robotics, AI, and embodied reasoning, offering a look at how generalist robots may handle useful real-world tasks.