Dynamic System Emulation: Fixed Wing Dynamics on a Multicopter

In the world of flight, there is a fundamental divide between how different aircraft move. Fixed-wing planes, like the ones that carry passengers across oceans, rely on forward speed to stay aloft. They cannot hover in place, and they must maintain a minimum velocity to avoid stalling and falling. Their movement is constrained by the laws of aerodynamics, making them efficient for long distances but difficult to control in tight spaces or for testing new ideas without risk. In contrast, multicopters, the familiar drones with multiple rotors, can fly in any direction, hover perfectly still, and take off vertically. They are incredibly agile but lack the specific flight characteristics of a plane. Engineers often need to test algorithms or train pilots on how a fixed-wing aircraft behaves, but using a real plane is expensive and dangerous. Virtual simulations exist, but they often fail to capture the messy reality of the physical world, creating a gap between computer models and actual flight.

To bridge this gap, researchers have developed a new way to make a drone behave exactly like a plane, not just in a computer, but in the real air. The core idea is to use a highly maneuverable multicopter equipped with a camera mounted on a two-axis gimbal—a mechanical mount that allows the camera to tilt and roll independently of the drone's body. The goal is to control the drone so that its camera sees the world from the exact same perspective as a fixed-wing aircraft would, even though the drone itself is flying in a completely different way. This approach, known as physical-to-physical emulation, offers a safe, low-cost platform for training and research that avoids the aerodynamic limits of real planes while providing real-world sensory data.

The researchers, working at Aarhus University, set out to prove that this mimicry is theoretically possible and to build a control system that could make it happen. They faced a significant challenge: the physics governing a plane and a drone are fundamentally different. A plane moves because of lift generated by its wings and thrust from its engines, while a drone moves by pushing air down with its rotors. A plane cannot simply stop and hover, but a drone can. To make the drone's camera follow the path and orientation of a plane, the researchers had to solve a complex mathematical problem of mapping the drone's movements to the plane's movements. They determined that by using a specific control technique called dynamic feedback linearization, they could transform the drone's complex, non-linear behavior into a simpler, linear form that could be precisely commanded. This method allowed them to calculate exactly how much thrust the drone needed and how the camera mount should move to perfectly match the viewpoint of a fixed-wing aircraft.

To test their system, the team ran detailed simulations using a model of a high-performance aerobatic plane, the Red Bull Edge 540, known for its ability to perform extreme maneuvers. They compared this against a model of a heavy-lift multicopter, the Freefly Alta-X, which has the power to carry the necessary equipment. The simulations put the system through two distinct tests. First, they had the drone mimic a simple, steady level flight. In this scenario, the drone had to tilt its body forward to generate the speed required to match the plane, but the camera mount had to rotate in the opposite direction to keep the camera level and pointing exactly where the plane's camera would be. The results showed that the system could track the plane's position and orientation with almost no error, keeping the camera view stable despite the drone's own tilting motion.

The second test was far more demanding. The researchers commanded the virtual plane to perform a vertical loop, a maneuver where the aircraft flies in a giant circle, going up, over the top, and back down. This creates rapid changes in speed, direction, and gravitational forces. During this aggressive maneuver, the drone had to tilt its body at extreme angles, exceeding 60 degrees at the top of the loop, to follow the curved path. While the drone's body twisted and turned violently, the camera mount worked in perfect opposition, rotating quickly to cancel out the drone's movements. The result was that the camera's view remained locked onto the exact path and orientation of the fixed-wing aircraft, even as the drone itself struggled to maintain the necessary thrust vector. The simulations confirmed that the system could handle these high-stress dynamics without losing stability or accuracy.

The findings demonstrate that a multicopter with a two-axis gimbal can successfully replicate the flight dynamics and camera viewpoint of a fixed-wing aircraft. The researchers proved that the drone's camera is fully controllable in all directions, allowing it to match the plane's position and attitude precisely. By using dynamic feedback linearization, they created a control framework that manages the complex relationship between the drone's body and the camera mount, ensuring that the camera sees the world exactly as the plane would. While these results were achieved in simulation, the work establishes a practical path toward building a physical system for pilot training, autonomy research, and testing new controllers. This approach offers a way to validate algorithms in the real world without the high costs and risks associated with flying expensive fixed-wing aircraft, potentially opening new doors for cross-system validation in aviation and robotics.