An autonomous glider, created by University of California San Diego researchers, uses machine learning to glide through the sky via atmospheric thermal plumes.
“This paper is an important step toward artificial intelligence—how to autonomously soar in constantly shifting thermals like a bird. I was surprised that relatively little learning was needed to achieve expert performance,” says Terry Sejnowski, member of the research team from the Salk Institute for Biological Studies and UC San Diego’s Division of Biological Sciences.
By employing this technique, scientists can further their understanding of bird migration. High-soaring birds seek warm, rising air passages known as thermals to stay aloft without flapping their wings.
To mimic this, researchers fitted their glider with an on-board flight controller, which “precisely controlled the bank angle and pitch, modulating these at intervals with the aim of gaining as much lift as possible,” according to the study.
On-board techniques also measure the glider’s roll-wise torques and local vertical wind accelerations.
The glider, which has a 6-ft (2-m) wingspan, teaches itself while on the move through “reinforcement learning.” The glider’s collective experiences over a period of several days accumulated into the aircraft’s complete navigational strategy.
In field experiments, the unmanned vehicle reached heights of nearly 2,300 ft (700 m).
“Our results highlight the role of vertical wind accelerations and roll-wise torques as effective mechanosensory cues for soaring birds and provide a navigational strategy that is directly applicable to the development of autonomous soaring vehicles,” according to the study.
The study, “Glider soaring via reinforcement learning in the field,” was published in Nature.
Filed Under: Aerospace + defense
