Race to use AI to respond to aircraft turbulence

According to the BBC, accidents caused by turbulence are increasing under the influence of human-caused climate change. Paul Williams, professor of atmospheric science at the University of Reading, UK, said that clear sky turbulence (CAT), a phenomenon of turbulent air flow that causes strong shaking in clear sky conditions and is difficult to detect with satellites, radar or the naked eye, has increased by 55% since 1979. It is expected that by the 2050s, air turbulence will triple globally, greatly affecting flight routes over East Asia and the North Atlantic.

In addition to its dangerous effects, air turbulence also damages the aviation industry, causing vehicle wear and tear and prolonging flight times, leading to more fuel consumption and increased emissions. Increasing turbulence in the atmosphere has airlines, scientists and engineers rushing to find ways to solve the problem.

The New York Post said startup Turbulence Solutions based in Vienna, Austria, is developing a small computer-controlled "flaplet" wing system that can be attached to an aircraft wing. The system determines wind conditions in real time and adjusts the wing angle accordingly to cope with changes in airflow, allowing the aircraft to move stably in turbulence.

According to Turbulence Solutions, their technology can reduce the shock felt by passengers by more than 80%. On small aircraft moving at speeds of more than 160 km/h that the company has tested, the wing adjustment time is only 0.002 seconds. Andras Galffy, the company's chief executive, said the flaplet system has proven effective on two-seat aircraft. He hopes to license the technology for small passenger aircraft by 2028 and large commercial aircraft by 2040.

Some companies like SkyPath are developing AI weather mapping software to help pilots before their planes go through turbulence. Most pilots carry iPads on planes, and SkyPath uses the device's built-in sensors to monitor vibrations and analyze atmospheric conditions. That information is fed into the AI ​​model along with more than 100 other meteorological parameters, helping to predict new disturbance events with 90% accuracy. Findings are displayed on a continuously updated map inside the app.

"We have billions of data points from aircraft and satellites, high-resolution atmospheric models and artificial intelligence models that combine it all into a comprehensive picture," said Dr. Williams, a SkyPath consultant.

Pilots use the data to circle over bumpy areas and help passengers be ready when turbulence becomes unavoidable. According to Maya Shpak, SkyPath's chief executive officer, the company has reduced the number of moderate turbulence encounters by more than 50%. SkyPath is partnering with many leading airlines globally including American, Southwest, Singapore and Emirates. To date, 130,000 pilots have signed up for the service and the company's platform is used on about 40,000 flights a day.

Ricardo Vinuesa, a researcher at the University of Michigan, USA, is applying AI to intelligently control airflow to prevent aircraft stalling and minimize sudden falls from high altitudes caused by air turbulence. Instead of spending a lot of time and effort on testing, Vinuesa's deep reinforcement learning AI model synthesizes millions of data points collected from wind tunnels over many years, then determines how different aircraft components react to drastically changing wind and weather conditions.

In early 2025, Vinuesa and colleagues at KTH Royal Institute of Technology, Barcelona Supercomputing Center and Delft University of Technology conducted experiments using AI to continuously observe the state of the air flow around the simulated aircraft wing in real time. Then, AI automatically makes accurate decisions when to spray air and with what frequency to optimize the efficiency of eliminating "turbulence separation bubbles". This is a phenomenon that occurs when the plane raises its head too high or the pressure changes, causing the air flow on the back of the wing to peel off, reducing lift and increasing drag, leading to loss of control (stall).

Flying directly into vortices and upward air masses requires not only precise engineering but also advanced mathematics and fluid dynamics analysis. Small disturbances from wind blowing around a building to another plane passing by can change the behavior of the air stream. AI can recognize those changes much easier than humans. "Machine learning is very good at finding rules in multidimensional data," BBC quoted Vinuesa. "Noise could be the perfect application for AI."