In order to let their superpressure balloons navigate through the stratosphere Loon and Google AI successfully use reinforcement learning (RL). This is a type of machine learning technique that enables an agent to learn by trial and error in an interactive environment using feedback from its own actions and experiences. This contrasts against the conventional approach of the automated system following fixed procedures artisanally crafted by engineers.
In a blog Loon explains why RL is practical for their fleet of stratospheric balloons. Loon’s navigation system’s most complex task is solved by an algorithm that is learned by a computer experimenting with balloon navigation in simulation. This is probably the world’s first deployment of reinforcement learning in a production aerospace system. The technical details of this achievement are now published in Nature.
Loon is a project to bring communication to remote areas using balloon, which in essence function like a floating network of cell towers, potentially powering billions of people who now have no connection to the internet. Read the blogpost here.