[ad_1]
The albatross is one of nature's most interesting creatures. They seem to fly for hours and hours and yet they only flap their wings on rare occasions. The bird achieves this wonderful feat, sometimes traveling 10,000 miles at a time, exploiting the dynamic skyrocketing, essentially capturing currents. It's hard to understand how birds climb in drafts, but we do.
The researchers reported in the newspaper Nature that they were able to design an artificial intelligence (AI) system that learned to take advantage of a particular type of airflow – hot air columns 39, elevating, called thermal – to fly a glider.
Credit: Pixabay.
AI used reinforcement learning, a type of machine learning where an agent learns to behave a environment by performing actions and seeing the results. In other words, the machine has not been educated on how to perform this task – he had to act optimally, maximizing the results bon a number of inputs. It's the same algorithm used by Google's infamous AlphaGo who learned to play the game. – a feat simply impossible with conventional programming techniques.
Right here, the machine was relayed by information such as glider pitch, yaw, ground speed and speed – settings that it was necessary to constantly tweak to reach the highest possible rate of rise.
Thermal ascents are essential to allow a glider to stay in the air as long as possible. In an updraft, increasing vertical air movement may be sufficient to stop the glider from falling and, if the vertical wind is strong enough, allow it to climb. Without an updraft, a glider will gradually fall to the ground.
Researchers at the University of California at San Diego and at the Abdus Salam International Center for Theoretical Physics in Trieste, Italy, first trained the machine learning algorithm in a simulator and then got their hands on the dough on the ground. They flew approximately 240 flights over Poway, Californiawho each lthree minutes on average. However, during some flights, the glider remained in the air for 45 minutes, flying at the same level as the eagles. The glider was even attacked bthere an eagle during a particular flight.
All the while, the AI was in full control, using thermals to get into the sky.
These early works suggest that the autonomous aircraft of the future could exploit the air currents to save energy. Such an unmanned aircraft could someday fly alongside migratory birds along the way, following each of their behaviors and generating important scientific knowledge.
Before that, however, there is still work to be done. The thermals are just one of the many types of air currents that a plane encounters in real life. A temperature current, or ascending air current, is formed when air masses of different temperatures meet. But there are also currents generated by the air coming off the mountains or by the collision of air movements in the convergence zones. Nevertheless, the work is impressive and promising.
It's also fascinating to see such powerful demonstrations that show how enhanced learning can be flexible – from glider control to eagles and Go World champions.
I liked this article? Join over 40,000 subscribers to the ZME Science newsletter. Subscribe now!
Source link
