Google launches new Gemini model for controlling robots without an internet connection.

 

A photo of Google DeepMind in Britain.

Google launches new Gemini model for controlling robots without an internet connection.

Google DeepMind, a subsidiary of Google, released a new language model on Tuesday called "Gemini Robotics On-Device," which can perform tasks locally on robots without an internet connection.

Building on the company's previous Gemini Robotics model released in March, "Gemini Robotics On-Device" can control robot movements.

Developers can control and fine-tune the model to meet different needs using natural language commands, according to a report by the technology news website TechCrunch, reviewed by Al Arabiya Business.

In performance tests, Google indicates that the model's performance is comparable to that of the cloud-based Gemini Robotics model.

The company notes that the new model outperforms other models running locally on devices in general performance tests, although it did not name those models.

In a demonstration, the company showed robots operating the new model performing tasks such as unzipping bags and folding clothes. Google notes that although the model was trained on ALOHA robots, it was later adapted to work on the dual-arm Franka FR3 robot and Apptronik's humanoid Apollo robot.

Google claims that its dual-arm Franka FR3 robot successfully handled scenarios and tasks it had never seen before, such as performing an assembly task on an industrial belt.

Other AI developers are also venturing into robotics. NVIDIA is creating a platform for creating basic AI models for humanoid robots.