Google’s AI division, Google DeepMind, has developed a new artificial intelligence that allows robots to self-train on new tasks by observing instructions on what needs to be done.
The AI is called “RoboCat,” and Google DeepMind states that it speeds up the learning process for robots and is a significant step towards creating a “general-purpose robot,” a robot capable of performing a wide range of tasks it is instructed to do.
RoboCat learns much faster than other state-of-the-art models. It can pick up a new task with as few as 100 demonstrations because it draws from a large and diverse dataset. This capability will help accelerate robotics research, as it reduces the need for human-supervised training, and is an important step towards creating a general-purpose robot.
–deepmind.com
In the video below, you can see how RoboCat is used to guide two different robotic arms equipped with RoboCat to perform various tasks they have learned through trial and error. Currently, these tasks are relatively simple, but it is not entirely impossible that RoboCat could serve as the foundation for more advanced robots in the future.
