Google DeepMind has introduced a cloud-free on-device VLA model for robotics, enhancing the autonomy and reliability of physical robots. This innovation allows robots to quickly adapt to their environments without relying on cloud processing, marking a significant advancement in generative AI applications in robotics.

The first beta release of OM1, an open-source and modular operating system for robots, has been announced, featuring integrations with multiple LLM providers, advanced autonomy capabilities, and simulator support. Key enhancements include support for various robots, speech-to-text and text-to-speech functionalities, and improvements in navigation and interaction with hardware components. Developers can leverage this release to prototype and deploy robotics applications across different platforms.