New Google AI makes robots smarter without the cloud

Google DeepMind has unveiled an on-device version of its Gemini Robotics AI, marking a significant shift in robotic autonomy. By processing advanced reasoning and control capabilities directly on the hardware, this system enables robots to execute complex tasks without relying on cloud connectivity. This innovation ensures fast, reliable performance in latency-sensitive environments, allowing intelligent operations even in areas with poor or no internet access. The model emphasizes adaptability and data privacy. It can be fine-tuned with minimal demonstrations and run entirely locally, keeping sensitive information secure within the device rather than transmitting it to external servers. This local-first approach is particularly relevant for open data initiatives, as it reduces the volume of telemetry data requiring centralized processing and storage. It supports developers in creating private, efficient robotic systems suitable for healthcare and infrastructure-poor settings where data sovereignty is critical. However, the architecture currently lacks built-in semantic safety features, placing the responsibility for safety implementation on developers. While this limits immediate broad deployment, it encourages transparent, customizable safety protocols aligned with specific operational needs. This move democratizes access to sophisticated robotic intelligence, allowing for more open, localized experimentation. Ultimately, it challenges the industry standard of cloud-dependent AI, proving that high-level autonomy can coexist with strict data privacy and offline functionality.

Source: foxnews.com
Published on 2025-07-03

New Google AI makes robots smarter without the cloud

Related news