American tech giant Google has released a new AI tool designed specifically for robotics. On Tuesday, the company announced Gemini Robotics On-Device, which can operate in a robot without needing a connection to the internet.
“Today, we’re introducing Gemini Robotics On-Device, our most powerful VLA model optimized to run locally on robotic devices”, the company revealed in a blog post. The new tool is a follow-up to Gemini Robotics which is a Visual Language Action (VLA) model introduced in March this year.
See Related: Jack Dorsey’s Block Announces DIY Bitcoin ‘Mining Development Kit’
Since it does not need to be online all the time, Gemini On-Device offers latency-sensitive applications and robustness in environments with poor connectivity. It is designed for bi-arm robots and requires little resources to operate. According to Google, this new AI delivers “strong visual, semantic and behavioral generalization across a wide range of testing scenarios”. In a video demo, the new AI is shown to be adept at tasks that require dexterity such as unlocking zip bags or folding clothes.
Additionally, Google has also released a Gemini Robotics SDK on GitHub. Developers can use the SDK (Software Development Kit) to assess the AI’s capabilities. Developers can also run the model on the MuJoCo physics simulator.