Learning-Based Robust Tracking Control of Quadrotor With Time-Varying and Coupling Uncertainties

Area of Science:

• Robotics and Control Systems
• Aerospace Engineering
• Artificial Intelligence

Background:

• Quadrotor unmanned aerial vehicles (UAVs) face complex dynamics with time-varying and coupling uncertainties.
• Existing control schemes struggle to guarantee robust tracking performance under such challenging conditions.
• External environmental disturbances further degrade the stability and accuracy of UAV systems.

Purpose of the Study:

• To propose a novel learning-based robust tracking control scheme for quadrotor UAVs.
• To address and compensate for time-varying, coupling uncertainties, and external disturbances.
• To provide theoretical stability guarantees for the proposed control strategy.

Main Methods:

• Modeling quadrotor dynamics incorporating uncertainties.
• Designing position and attitude tracking error subsystems.
• Employing neural networks and an improved weight updating rule for approximate optimal control.
• Utilizing learning-based approaches to derive control laws for nominal error subsystems.

Main Results:

• The proposed scheme demonstrates robust tracking control despite system uncertainties.
• Theoretical stability of tracking error subsystems is established.
• Simulations on linear and nonlinear quadrotor models show competitive and effective performance against variable disturbances.

Conclusions:

• The developed learning-based robust tracking control scheme is effective for quadrotor UAVs.
• The approach provides a reliable method for achieving stable and accurate flight in uncertain environments.
• The findings validate the scheme's potential for real-world applications facing dynamic challenges.