Alphabet's Loon said its balloons are using a self-learning AI system for navigation, according to online publication Engadget.com. The system was created by Loon with Google AI over the last few years and can figure out the optimal route for balloons faster than the balloons' previous human-made navigation system. The system was designed together with Google's AI team in Montreal, TheVerge.com said.
A fleet of balloons is now using the reinforcement learning (RL) flight control system over Kenya, where Loon launched its first commercial internet service in July.
Loon says this is the first use of an RL model in "a production aerospace system." Loon and Google AI used simulations to train the RL model through trial and error before real-world testing began in Peru. In the Peru trial, the AI-controlled system outperformed the human one by staying closer to a device the team uses to measure LTE signals in the field.
Every day we send out a free e-mail with the most important headlines of the last 24 hours.
Subscribe now
We welcome comments that add value to the discussion. We attempt to block comments that use offensive language or appear to be spam, and our editors frequently review the comments to ensure they are appropriate. If you see a comment that you believe is inappropriate to the discussion, you can bring it to our attention by using the report abuse links. As the comments are written and submitted by visitors of the Telecompaper website, they in no way represent the opinion of Telecompaper.