June 24, 2025:
Google Launches On-Device Gemini Robotics Model - Google DeepMind has introduced the Gemini Robotics On-Device model, allowing robots to perform tasks without internet access. This advancement of the March-released Gemini Robotics model lets developers control robotic movements through natural language prompts. Demonstrations have shown robots completing tasks like unzipping bags and assembly on industrial belts.
Alongside this, Google is launching a Gemini Robotics SDK for training robots using the MuJoCo simulator. Competitors such as Nvidia and Hugging Face are also making strides in robotics with their foundation models.
