Control of UAV quadrotor using reinforcement learning and robust controller

Abstract

The control of unmanned aerial vehicle quadrotor is challenging because of non‐linearities, coupling and disturbance. Here, a novel control method which includes reinforcement learning (RL) component and robust component is proposed. In this method, the RL component only relies on collected data instead of modelling to handle coupling and disturbance from aerodynamics and model. To ensure safety during training and improve training speed, the robust component is used to reduce the disturbance .The stability of the system with our controller is proven by Lyapunov method. The results of the simulation exhibit advanced performance of our controller.

A Model-Free Online Learning Control for Attitude Tracking of Quadrotors

“…This includes offline learning methods such as reinforcement learning and deep reinforcement learning. These methods learn control policies directly from flight data and thus avoid having to use a UAV model [14,15]. However, their generalization remains difficult to ensure, and their performance and stability should be further verified when the scene that generates the data for training differs greatly from the actual scene.…”