When you purchase through links on our articles, Future and its syndication partners may earn a commission.

Scientists have developed a technique that could mitigate the effects of turbulence on dynamic structures and vehicles, with a particular focus on unmanned aerial vehicles (UAVs).

Turbulence is the name we give to changes in air pressure that cause planes to shake. This is particularly evident when an aircraft shakes when experiencing changes in air pressure during flight. This is unlike flying animals, which have evolved a natural ability to detect changes in their environment that cause turbulence and quickly adapt to maintain smooth flight.

Research published September 24 in the journal NPJ Roboticsexplained how scientists could develop a technique for controlling aircraft. The technique required the use of a artificial intelligence (AI) called FALCON to automatically adjust flight to compensate for turbulence.

Reinforcement learning – an AI training method – has already been used to develop AI-augmented control systems, but only for specific environments or vehicles. FALCON, on the other hand, has been trained to understand the underlying principles behind turbulence in order to adapt to all conditions.

FALCON is based on Fourier methods, which use complex sine waves to represent data. Researchers found that digitally representing wind conditions as periodic waves was an effective way to model turbulence because the ebb and flow of wind and its effects naturally follow a wave pattern.

Related: New “wastewater” jet fuel could reduce aircraft emissions by 70%

“Using reinforcement learning to adapt in real time is remarkable because it learns the underlying turbulence pattern,” Hever Moncayoprofessor of aerospace engineering at Embry-Riddle Aeronautical University told Live Science. “I think this technology is very feasible, especially with current computing capabilities such as Jetsonwhich support real-time integration of adaptive learning, Fourier analysis and computation.

The scientists tested the AI ​​in a wind tunnel at Caltech, using an aerodynamic wing to represent a drone and equipping it with pressure sensors and control surfaces. He used them to sense changes in pressure and adjust his tilt and yaw as necessary to maintain stability. A movable cylinder was also placed upstream of the wing in the wind tunnel to generate random turbulence fluctuations.

It was found that after nine minutes of learning, during which FALCON continually attempted to adapt to changes in turbulence and return results, the AI ​​could maintain the stability of the canopy in the wind tunnel.

“Caltech’s wind tunnel tests show that FALCON can learn in minutes, indicating scalability to larger aircraft,” Moncayo said. “However, practical challenges remain, particularly in quickly adapting to diverse and unpredictable conditions and in validating performance in various drone configurations and wind environments.”

RELATED STORIES

—Largest all-electric concept plane ever could take flight by 2033

—Hydrogen-powered VTOL plane completes record 523-mile journey – and lands with 10% of its fuel in the tank

—New in-car AI algorithm can spot drunk drivers by constantly scanning their faces for signs of intoxication

By enabling automated adaptation to turbulence, this research could potentially lead to smoother flight for drones and commercial aircraft. Researchers have also suggested the possibility of sharing environmental data between planes to prevent disruptions. However, given the cybersecurity concerns surrounding aircraft control systems, this would require a robust security protocol that would need to be carefully reviewed and tested beforehand.

“Further development will likely focus on improving forecast accuracy and reducing training time, which is achievable but complex,” Moncayo said. “Additionally, sharing information between aircraft will improve the predictive power of the system, but will likely require robust communications standards and data processing protocols for wider adoption.”

The next stage of research aims to reduce the training time of AI. This will likely become the main challenge for researchers, as the ability to quickly adapt to environmental conditions is essential to finding a practical solution to turbulence.