It’s challenging to picture a drone replicating the remarkable skills of a flying squirrel—such as gliding, adjusting course in midair, and navigating forest canopies with exceptional agility. However, this is exactly what inspired a novel type of airborne robot, featuring adaptable, collapsible wings along with an ‘intelligence’ system driven by artificial intelligence algorithms.
Engineers in South Korea have created a drone that replicates such aerial maneuvers — a quadrotor fitted with folding wing membranes capable of abruptly decelerating, making acute turns, and dodging obstructions in manners unachievable for conventional drones.
The team responsible for the innovative design, which involves a partnership between POSTECH and ADD’s AI Autonomy Technology Center, aims to enhance drones’ ability to maneuver through confined or uncertain spaces such as forests, disaster areas, or city landscapes with tall buildings.
From Woods to Flight Labs
Open the Youtube video
Flying squirrels do not truly fly; instead, they glide. By extending membranes that connect their wrists to their ankle bones, these creatures form makeshift wingsuits. This adaptation enables them to navigate intricate landscapes efficiently and importantly, to slow down significantly right before touching down.
These features make flying squirrels extremely agile, which can’t be said about your typical quadcopter. The same features that make them stable — fixed rotors and rigid frames — also limit how sharply they can turn or respond to sudden obstacles.
To make quadcopters more squirrel-like, the South Korean researchers designed feather-light silicone wings — just 24 grams in weight — that can fold and unfold with servo motors.
However, wings are not sufficient on their own. The true enchantment comes from synchronizing them effectively.
This is precisely what the Thrust-Wing Coordination Control system, known as TWCC, addresses. The system continuously evaluates if extending the wings will aid or impede the drone’s navigation. If the internal control unit forecasts that an action might surpass the drone’s pitch or roll thresholds, it triggers the wing deployment, thereby increasing the force potential without causing unstable conditions for the aircraft.
It utilizes a series of sensors including GNSS, barometers, and inertial measurement units to monitor its location and alignment. This data is processed by the TWCC algorithm, which instantly determines when to extend or retract the wings, enabling the drone to maneuver swiftly, change direction, or halt as required.
“The wings are extended, and thrust is calibrated… enabling [the drone] to produce a more powerful force in the intended direction,” the team detailed in their study, which has been uploaded as a preprint on the server.
arXiv
.
The drone does not depend on a powerful supercomputer or remote servers; instead, it functions independently using just a basic microcontroller unit (MCU). This type of MCU can be found in DIY electronic projects similar to those involving Arduino boards. The ability to run advanced artificial intelligence on an inexpensive, readily available chip is among the most remarkable aspects of this initiative.
Avoiding Obstacles at Speed
To determine whether the drone was capable of navigating actual chaotic environments, the group put it through tests outside using a course designed with simulated barriers. Whenever this particular typeless drone got close to an impediment, it had difficulty—either deviating from its route or dropping height significantly. Its engines were not powerful enough to provide sufficient upward thrust when performing abrupt movements.
However, the flying squirrel drone, employing both propeller propulsion and lift from its wings, sustained its route and elevation throughout the intricate obstacle course. It demonstrated efficient climbing and braking capabilities with negligible drifting. During an experiment, it navigated bends at 7.3 meters per second (approximately 26 km/h), managing to enhance its path accuracy by almost a meter over the wingless variant.
“The wing membranes’ effect in the actual experiment seems to surpass what was seen in the simulation,” the team remarked.
This additional lift and drag reduced pressure on the batteries as well. By decreasing the load on the engines when performing abrupt movements, the drone steered clear of brownouts and instability issues common with conventional quadcopters.
This is not the initial effort to incorporate passive surfaces into drones, but it could represent the most comprehensive fusion of flexible designs, physics-driven learning algorithms, and instantaneous control mechanisms. Moreover, this breakthrough paves the way for innovative uses.
In areas affected by disasters, such drones could maneuver through narrow spaces filled with rubble. Within forests, they have the potential to track animal movements. Additionally, equipped with enhanced collision prevention features, they could operate securely in city landscapes or crowded warehouse corridors.
The authors are already looking ahead. Future versions could include even more advanced trajectory planning, allowing the drone to anticipate not just the next move — but the smartest one.
This story originally appeared on
ZME Science
. Looking to become smarter each day?
Subscribe to our newsletter
and keep up-to-date with the newest scientific discoveries.