Model-Free Attitude Control of Quadcopter using Disturbance Observer and Integral Reinforcement Learning

A model-free attitude controller is designed for quadcopter systems using extended state observer and integral reinforcement learning. The extended state observer enables controller design even without the complete knowledge of system dynamic model. As a baseline controller, an integral reinforcement learning approach is employed, which is updated with online data along with system trajectories. Numerical simulations demonstrate the effectiveness and robustness of the proposed method for quadcopter attitude control.