<ul data-eligibleForWebStory="true"><li>Google DeepMind introduced the Gemini Robotics On-Device model for on-device robotics tasks, offering adaptability and accessibility.</li><li>The on-device model is designed to bring multimodal reasoning and understanding into the physical world, optimized for two-armed robots with minimal computational requirements.</li><li>DeepMind also released the Gemini Robotics SDK for developers to evaluate and adapt the model for various tasks and environments, emphasizing low-latency inference and robustness.</li><li>Gemini Robotics On-Device demonstrates strong generalization in visual, semantic, and behavioral aspects, enabling robots to perform complex tasks like following natural language instructions and executing dexterous manipulations.</li></ul>

Google DeepMind introduces on-device Gemini AI model for robots

Discover more