Air turbulence affecting aircraft may soon be a thing of the past, thanks to a new AI system that enables flying vehicles to learn how to adjust to turbulence in just a few minutes.
Scientists from Embry-Riddle Aeronautical University (USA) have developed a technique that can minimize the impact of turbulence on flying vehicles, particularly unmanned aerial vehicles (UAVs).
Air turbulence is a terrible experience when flying – (Photo: REUTERS).
This technique relies on a machine learning system called FALCON that adjusts flight behavior to adapt to external turbulence.
Air turbulence refers to changes in air pressure that cause aircraft to shake. FALCON is trained to understand the fundamental principles causing turbulence so it can adapt under any conditions.
This AI system is based on Fourier methods, which utilize complex sine waves to represent data. According to LiveScience on November 11, the research team tested this AI system in a wind tunnel at the California Institute of Technology (Caltech, USA), using a wing equipped with pressure sensors representing a UAV. FALCON uses these sensors to detect pressure changes and adjusts altitude and deflection as needed to maintain stability.
The team found that after 9 minutes of learning and continuously trying to adapt to changing turbulence, FALCON was capable of maintaining the stability of the wing in the wind tunnel.
“Wind tunnel tests at Caltech show that FALCON can learn in just a few minutes, with the goal of scaling it up for larger aircraft,” said Professor Hever Moncayo, who works at Embry-Riddle University.
By enabling automatic adaptation to turbulence, this research has the potential to help UAVs and commercial aircraft fly more smoothly in the future. The research team also proposed the possibility of sharing environmental data among aircraft to alert them to turbulence.
The next phase of the research aims to reduce FALCON’s learning time. This could be a significant challenge for the team, as the ability to quickly adapt to environmental conditions is crucial for practical solutions to turbulence.
Additionally, there are still other challenges in the real world, especially due to the diverse and unpredictable nature of wind conditions.
The research was published in the journal NPJ Robotics.
