We always thought Project Loon would operate by creating rings of balloons around the globe and worked hard to maximize the amount of time each balloon spent over land. But over the last few months we were able to make some major breakthroughs by applying machine learning techniques to our navigation algorithms, and as a result, we can now get teams of balloons to cluster together over a particular region of the earth. This has enormous implications for Loon’s economic and operational viability: we can put together a Loon network over a particular region in weeks not months and we can get greater value out of each individual balloon. Overall this means we can reduce the total number of balloons we’d need, which has huge implications for the costs of operating a Loon-powered network.
64 plus ones
Shared publicly•View activity
- That's awesome!19w
- Congrats!! Wow, 100x instead of the usual 10x! Unbelievable! Go machine learning!19w
- LOL YES. Now when they go prioritize someone else's connection for a while I can send out my drone to tow a balloon over to my house hahahahahahahahaha19w
- Around the US, it is approx 10km into the air. Good luck getting a drone that far up.19w
- It'll just have to be a beastly drone.18w