SoylentNews

taylorvich [soylentnews.org] writes:

https://techxplore.com/news/2025-04-flying-squirrel-drone-foldable-wings.html [techxplore.com]

Unmanned aerial vehicles (UAVs), commonly known as drones, have already proved to be valuable tools for a wide range of applications, ranging from film and entertainment production to defense and security, agriculture, logistics, construction and environmental monitoring. While these technologies are already widely used in many countries worldwide, engineers have been trying to enhance their capabilities further so that they can be used to tackle even more complex problems.

Researchers at Pohang University of Science and Technology and the Agency for Defense Development (ADD)'s AI Autonomy Technology Center in South Korea recently developed a drone with foldable wings that could be more maneuverable than conventional drones. Their drone draws inspiration from the winged flying squirrel, a type of squirrel that uses loose flaps of skin attached from their wrists to their ankles to glide from tree to tree.

"The flying squirrel drone is inspired by the movements of flying squirrels, particularly their ability to rapidly decelerate by spreading their wings just before landing on trees," Dohyeon Lee, Jun-Gill Kang and Soohee Han, co-authors of the paper, told Tech Xplore. "We initiated this research with the belief that, like flying squirrels, drones could expand their dynamic capabilities by utilizing aerodynamic drag."

Published on the arXiv preprint server, the most recent paper by Lee, Kang and Han builds on an earlier paper in which they first presented their squirrel-inspired robot. Their previous paper outlined their robot's basic underlying hardware and a reinforcement learning technique that allowed it to rapidly decelerate while performing maneuvers in flight.

"In our new paper, we proposed a novel drone system that utilizes deployable wing membranes, demonstrating superior performance compared to conventional drone systems in executing high-acceleration maneuvers such as rapid stops and sharp turns," said Lee, Kang and Han.

The aerodynamic drag from the robot's wing membrane was previously found to hinder its performance in common flight scenarios (i.e., when the drone was flying in a straight line). In situations where the robot is required to suddenly stop or rapidly change direction to prevent collisions with obstacles, the deployment of wings can produce a sizeable force in the direction opposite to the object that the drone is trying to avoid.

"To safely and reliably operate in these scenarios, the flying squirrel drone must be capable of deciding when to deploy or retract its wings based on the situation, and the rotors must be able to generate appropriate thrust accordingly," explained Lee, Kang and Han.

In their recent study, the researchers also trained artificial neural networks to accurately predict the aerodynamic drag generated by the drone's silicone-based wing membrane. They then developed a Thrust-Wing Coordination Control (TWCC) strategy that utilizes the neural network's predictions to optimally control both the wing membrane and motors, enabling the reliable execution of desired maneuvers.

"Another key contribution of our work is the development of a hardware system that allows for rapid deployment and retraction of the silicone wings, all while maintaining the conventional quadrotor form factor," said Lee, Kang and Han.

"Collectively, we proposed a framework capable of simultaneously controlling the silicone wing membrane—with its complex, non-analytically predictable aerodynamics—and the drone's motors, and the demonstration of high-performance trajectory tracking and obstacle avoidance on real hardware."

Original Submission