RoboCat, DeepMind’s robot cat, can learn to perform robotics tasks
DeepMind claims to have developed an AI called RoboCat that is capable of performing a variety of tasks on different robotic arms. This is not particularly novel. DeepMind says that this model is the world’s first to solve multiple tasks, adapt to them and do it using real robots.
In an email interview with TechCrunch, Alex Lee, DeepMind’s research scientist and co-contributor to the team behind RoboCat said: \”We demonstrate that one large model can solve diverse tasks on multiple robotic embodiments, and can adapt quickly to new tasks and embodiments.\”
RoboCat, which was inspired from Gato, DeepMind’s AI model capable of analyzing and acting on text, images, and events, was trained using images and action data collected in both simulation and real-life robotics. Lee says the data came from a mixture of other robot-controlling model inside virtual environments, humans operating robots, and previous iterations RoboCat.
Source:
DeepMind’s RoboCat learns to perform a range of robotics tasks
