Experimental Validation of Fuzzy Geometric Control for Quadrotor Attitude Control on AR-Drone 2.0

Abstract

Reliable attitude control is essential for quadrotor UAVs deployed in critical applications such as surveillance and delivery. Conventional PID, geometric, and quaternion-based controllers face limitations when handling large-angle rotations, singularities, and real-world disturbances. This paper presents a Fuzzy Geometric Control (FGC) framework directly formulated on the Special Orthogonal Group SO(3), eliminating small-angle approximations and quaternion ambiguities while enabling full-attitude maneuvers. By integrating fuzzy inference with geometric control, the proposed FGC achieves enhanced robustness against modeling uncertainties and indoor/outdoor disturbances. The controller is implemented in real time on a commercial AR.Drone 2.0, replacing its internal control unit to validate the approach on a practical platform. Experimental results show precise tracking with RMSE values of 2.1° for roll, 2.7° for pitch, and 7.2° for yaw under internal and external perturbations, minimal overshoot (<2%), and fast recovery times (<3.7 s). Compared to the baseline PID and the classical geometric controller, the proposed FGC reduces yaw overshoot from 39.9% to 6.18% (≈84% reduction), decreases yaw settling time from 6.25 s to 2.1 s (≈66% faster convergence), and lowers typical attitude RMSEs by ∼25–30% across axes. These findings confirm that the FGC scheme is both theoretically consistent and practically effective, providing a reliable control solution for UAV operations in complex environments.