Whoever thought balloons can be smart enough to catch winds for a free ride to their destination.
When Google’s Project Loon was launched, the balloons used cleverly coded static algorithms to maintain its altitude and position despite changes in weather conditions.
Although clever, these balloons while providing internet services could not adapt to unexpected changes in weather patterns.
Given their altitude of tens of thousands of feet, it is critical that they remain more or less static despite dynamic changes in weather patterns for a prolonged period of time.
In order to factor in changes in unexpected weather, Google’s Loon team has now revealed that it is using artificial intelligence in the form of machine learning to adjust the balloon’s behaviour high up in the air, with astonishing results.
One test balloon has stayed up in the Peruvian stratosphere for a mind blowing 98 days. It did so by adapting to changing wind conditions, which at those heights could have easily drifted it away.
Thanks to the Loon team’s smart coders, the balloon’s algorithms can now sift through large quantities of data and learn from it. Case in point: while one balloon floated away to the Pacific Ocean, it determined which gust of wind to ride to find its way back to land.
The balloon’s algorithms also use “reinforcement learning,” wherein it refines its behaviour from its behavioral predictions. As a result of these changes in its algorithm, a test balloon made close to 20,000 tweaks to maintain its altitude and course in the last 14 weeks.
The next AI upgrade should allow them to be afloat for even longer which could help trim their cost of staying afloat. Furthermore, the number of balloons required to cover an area for internet access could also come down.
This major development could be crucial for those who need internet access in remote places, which if not for these balloons, would be an impossibility.