Google DeepMind reveals new AI robot control models

Google Deepmind, AI Research Lab on Google, Wednesday Announced new AI models called Gemini Robotics Designed to enable machines in the real world to interact with objects, to navigate the environment and others.

Deepmind has published a series of demonstration videos showing robots equipped with folding paper on the robotics of Gemini, putting a pair of glasses in a case and other tasks in response to voice commands. According to the laboratory, Gemini Robotics is trained to summarize behavior in a number of different robotics hardware and to connect elements that robots can “see” with actions they can take.

Deepmind claims that in tests, Gemini robotics allows robots to perform well in environments not included in training data. The laboratory has launched a weakened model, the Gemini Robotics-Er, which researchers can use to train their own robotics control models, as well as a benchmark called Asimov to measure risks with AI-nourished robots.