Control Method Helps Drones Stay Stable in Turbulent Conditions

Unmanned aerial vehicles, especially quadrotors, are increasingly relied upon for tasks ranging from surveillance to delivery. Yet maintaining their stability in unpredictable environments remains a fundamental challenge. External disturbances, modeling errors, and noise can destabilize control systems, leading to degraded performance or even failure. Addressing these challenges requires control strategies that can adapt in real time while ensuring reliable and bounded system behavior.

Professor Francisco Jurado from Tecnológico Nacional de México/Instituto Tecnológico de La Laguna developed a new adaptive control framework designed specifically for quadrotor pose stabilization. His work, published in the peer-reviewed journal Applied Sciences, introduces a decentralized robust model reference adaptive control approach, a control method that continuously adjusts itself to follow a desired behavior, enhanced through a technique known as e-modification. As Professor Jurado explains, “In this work, a decentralized robust direct model reference adaptive controller via e-modification is suggested for the pose control of a quadrotor to prevent parameter drift.”

Professor Jurado focused on one of the most persistent issues in adaptive control systems: parameter drift, a situation where estimated values move away from their true/ideal values over time. This phenomenon occurs when estimated parameters deviate significantly from their true/ideal values, especially under disturbances or insufficient excitation signals, meaning the system does not receive enough varied input to learn properly. In practical terms, this can cause sudden divergence in system output, making drones unreliable in real-world conditions. The proposed method introduces an error-dependent damping mechanism, a stabilizing adjustment that depends on how large the tracking error is, that dynamically adjusts parameter updates, preventing instability while maintaining responsiveness.

Simulation results demonstrate that the new control strategy performs effectively across multiple scenarios. When disturbances such as noise are absent, the system parameters converge smoothly toward their ideal values, meaning the controller learns the correct behavior over time. Under more realistic conditions, including external perturbations, the controller maintains stable tracking performance without allowing parameters to diverge. Instead of drifting uncontrollably, the system remains within bounded limits, meaning its behavior stays within safe and predictable ranges, ensuring reliable operation. This balance between adaptability and robustness marks a significant improvement over conventional adaptive control methods.

A key strength of the approach lies in its decentralized design. Rather than treating the quadrotor as a single complex system, the rotational dynamics, the motions describing how the drone tilts and turns in space, are divided into smaller subsystems corresponding to roll, pitch, and yaw. Each subsystem is controlled independently while still accounting for their intrinsic coupling, meaning the motions still influence each other. This structure simplifies the control design and enhances robustness, particularly when dealing with uncertainties that affect different axes differently.

Professor Jurado’s study also compares the proposed e-modification approach with other established techniques, including sigma-based modification and smooth dead-zone adjustment method. The smooth dead-zone method temporarily stops adaptation when errors are very small to avoid unnecessary changes. While all methods aim to mitigate parameter drift, the results show that the e-modification strategy achieves consistently lower tracking errors in most cases. At the same time, it avoids undesirable effects such as oscillations or degraded performance that can arise with alternative methods. Importantly, Professor Jurado confirms that “Uniform ultimate boundedness of the tracking error signal is ensured.” This means that, over time, the tracking error will remain within a fixed safe range even if disturbances are present.

Beyond quadrotors, the implications of this work extend to other aerospace and mechanical systems where precise attitude control, the ability to control orientation in space, is essential. The ability to maintain stability despite uncertainties is particularly valuable in applications such as satellite orientation, where disturbances are small but persistent. By combining adaptability with mathematical guarantees of bounded behavior, the proposed method offers a practical pathway toward more reliable autonomous systems.

As Professor Jurado emphasizes, “Although for most of the cases the control community is just interested in the achievement of the control task by the adaptive controller, without paying attention if the parameter estimates converge to their true/ideal values, the parameter drift can not be neglected.”

In summary, Professor Jurado’s research presents a robust and efficient solution to a longstanding challenge in adaptive control. By preventing parameter drift and ensuring stable tracking, the decentralized model reference adaptive control approach with e-modification enhances the reliability of quadrotor systems operating in uncertain environments. The findings not only advance control theory but also contribute to the safe and effective deployment of drones in increasingly complex real-world scenarios.

Journal Reference

Jurado F., Ollervides-Vazquez E.J. “Decentralized Robust Direct MRAC via e-Modification for the Pose of a Quadrotor UAV.” Applied Sciences, 2025; 15: 11713. DOI: https://doi.org/10.3390/app152111713

Image Reference

Quadrotor under strong wind disturbances. AI-generated image created with ChatGPT (DALL E, OpenAI), 2026. The figure illustrates a quadrotor maintaining flight under significant turbulence and strong wind gusts, with visible air flow perturbations around the rotors.

About the Author