Birds have long inspired humans to fly. It took mankind centuries to learn and find their own way to fly. Now artificial intelligence (AI) learn to do same but in a short span of time. No, it’s not a plane. It’s an artificial intelligence glider that learns to fly in the same way that birds do.
Birds don’t always go from A to B by flapping their wings. There are soaring bird species that migrate long distances by using thermal updrafts technique. Soaring birds use warm air passages known as thermals to fly and gain height without needing to flap their wings, although no one knows exactly how they do it.
To gain insight into what cues birds instinctively use to catch an updraught, scientists from the University of California San Diego build this self-flying glider.
The aim of this self-flying glider is to learn every avian trick of birds and then apply that knowledge to create pilotless glider with an onboard computer allowing it to change direction on the basis of real-time measurements.
The team reported that after just 15 hours of test flights, the glider had “learned” how to optimize its position and developed a strategy to catch the warm updraughts.
As described in a paper published this week in the journal Nature, researchers used machine learning to train an algorithm in flight simulators. The gliders were trained using reinforcement learning, a type of machine learning where a software agent learns how to behave by performing actions and seeing the result.
The system is given a number of inputs consisted of specialized instruments capable of reading the change in upwards (vertical) wind strength. The instruments were able to identify the changes along the length of the glider (longitudinally) and from one wing tip to the other (literally). The sensors were able to make these measurements ten times every second.
This data was then used to make flight adjustments to what is known as the bank angle of the of the glider. A well-balanced airplane with its wings level has zero bank angle and will fly in a straight line. Tilting the wings and increasing the bank angle will make the plane turn.
Two example plots of the glider’s flight data. The green dot is its starting location, and the red dot is its finishing point. Image: Gautam, Wong-Ng, Celani, Sejnowski, VergassolaIn the study, the glider was rewarded if the change in upward wind speed along its flight path increased. In other words, if the glider was flying into an updraft.
By finding updrafts which help it stay aloft, the glider can slip through the air indefinitely, much like birds to when trying to minimize their energy output. It starts without any knowledge of the task, and it learns how to behave correctly over time.
As a bonus, the researchers used a numerical model to show this approach would benefit larger gliders even more since their longer wingspan will provide a more accurate measurement of the change in upwards wind speed from one wing tip to another.
After a number of flights, about 16 hours of flying in total, the study glider learned to fly by training itself that under a certain combination of inputs (bank angle, longitudinal and lateral change vertical wind speed) to decide what the next change in bank angle should be.
The researchers trained their algorithm first in a simulator and then in real life. They performed some 240 flights in the skies over Poway, California, which lasted, on average, about three minutes. They steered their glider to a fixed location using a manual controller, and then the AI took over, using the air currents from thermals (which can travel as fast several meters a second) to climb into the sky.
The result was that by the end of all that flying the plane had taught itself how to fly into updrafts, allowing it to stay in the air for longer.
Thought the result was impressive there are lots of things that needed to fix. Thermals are just one type of updraft that soaring birds take advantage of. Others are created by air currents spilling over mountain ridges or by the collision of air masses. In other words: just because AI can ride a thermal, doesn’t mean it’s ready to take on the varieties of wind that the world has to offer.
“This paper is an important step toward artificial intelligence—how to autonomously soar in constantly shifting thermals like a bird,” said Terry Sejnowski, a member of the research team from the Salk Institute for Biological Studies.
“I was surprised that relatively little learning was needed to achieve expert performance.”
More in AI :