Google DeepMind has unveiled a new robotics AI model demonstrating advanced reasoning capabilities in general-purpose machines. The AI model enables robots to perform complex tasks such as sorting laundry, unzipping bags, pouring liquids, and even tying shoelaces. This new on-device AI model can operate without an internet connection, marking a significant advancement in autonomous robot control and adaptability. The AI model allows robots to adjust to new situations, comprehend and respond to commands, while also performing tasks that require nuanced motor skills.
This technology signifies a push to make robots more useful in everyday scenarios. The same AI model can now load a dishwasher, after being fed new data, showcasing its adaptability. It can isolate stacked plates, pick up glasses, and adjust to different starting configurations, recovering from errors. This eliminates the need for new algorithms or special-case engineering, as the model learns from new data.
These advancements have the potential to increase efficiency, safety and commercial gains. Companies are exploring ways to incorporate robotics and automation into their operations. The model can run locally on robots without needing an internet connection.
