Google DeepMind has launched the Gemini Robotics On-Device model, enabling robots to operate autonomously without cloud connectivity. This innovation enhances real-world adaptability and responsiveness, particularly in environments with limited or no internet access. By incorporating multimodal reasoning, the model can understand natural language and perform tasks such as unzipping bags or folding clothes with minimal demonstrations. Developers can access the model through Google’s tester program and utilize a software development kit (SDK) for customization. Unlike its cloud-dependent predecessor, Gemini Robotics On-Device runs entirely on the robot, allowing for faster responses and better reliability. While it lacks built-in semantic safety reasoning, Google encourages developers to establish their own safety protocols. This on-device version is designed for privacy-sensitive applications, as it processes data locally. Overall, this launch marks a significant advancement in the development of adaptive and secure robotics for various practical applications.
Source link
