Attitude Stabilization of a Tricopter UAV Using Integral-Augmented Linear Quadratic Regulation

Abstract

Attitude stabilization of a tricopter unmanned aerial vehicle (UAV) is challenging because its asymmetric three-rotor configuration and servo-actuated rear rotor introduce coupled and disturbance-sensitive rotational dynamics. This study develops and experimentally evaluates an integral-augmented Linear Quadratic Regulator (LQR-I) for roll, pitch, and yaw stabilization of a physical tricopter UAV. The controller was implemented on an embedded flight controller using inertial measurements from an MPU6050 and was evaluated through tuning-rig experiments and outdoor free-flight tests under measured wind disturbances. A conventional LQR was used as the baseline controller. To ensure a consistent comparison, the performance of both controllers was evaluated under the same measured wind speed of 4.4 m/s. Under this condition, LQR-I reduced pitch overshoot from 9.03° to 2.85° and yaw overshoot from 13.43° to 3.35°, corresponding to reductions of approximately 68.4% and 75.1%, respectively. Roll overshoot increased slightly from 5.25° to 5.47°. The magnitude of the steady-state error was reduced by approximately 56.6%, 16.6%, and 50.6% for the roll, pitch, and yaw axes, respectively. In addition, the root mean square error (RMSE) decreased from 2.451° to 2.065° for roll, from 3.397° to 2.542° for pitch, and from 2.744° to 1.227° for yaw. These experimental results demonstrate that integral augmentation improves the overall disturbance-rejection and attitude-tracking performance of the tricopter, particularly in the pitch and yaw axes, although a slight increase in roll overshoot was observed.