Google DeepMind has introduced an on-device version of its Gemini Robotics AI model, allowing operation without an internet connection. This vision-language-action model retains dexterous capabilities akin to the original version, enabling robots to perform various physical tasks effectively. Users benefit from offline functionality and adaptability with minimal training. While the flagship model offers superior performance, the on-device variant serves well in scenarios with poor connectivity. Google is also launching an SDK for developers to fine-tune the model, catering to different types of robots and security needs.
The flagship Gemini Robotics model is designed to help robots complete a wide range of physical tasks, even if it hasn't been specifically trained on them.
With the device-only model, users can access offline features that are almost as good as those of the flagship.
I would think about it as a starter model or as a model for applications that just have poor connectivity.
The on-device Gemini Robotics model and its SDK will be available to a group of trusted testers while Google continues to work.
Collection
[
|
...
]