Robust Visual Servoing for UAVs Accounting for Dynamic Parameter Uncertainties

This paper presents a robust adaptive visual servoing strategy for quadrotors with unknown mass-inertial parameters and external disturbances. The proposed framework utilizes an onboard monocular camera and IMU to regulate the quadrotor’s 3D position and yaw relative to a planar target. A dual-loop control structure is developed: the outer-loop integrates backstepping-based visual servo control with virtual image moments, while the inner-loop employs a simplified torque controller using linear regression for inertia estimation. Sliding mode observers and adaptive laws are combined to compensate for time-varying disturbances and parametric uncertainties, avoiding depth dependency. Simulations confirm robustness against uncertainties and bounded tracking errors. The work advances practical vision-based quadrotor control in uncertain environments.