A reinforcement learning method of PID parameters in UAV’s flight control

In order to improve the dynamic performance and steady-state precision of the flight controlling system of Unmanned Aerial Vehicle (UAV) in complex environments, this paper proposes using the Proportion-Integration-Differentiation (PID) controller based on Reinforcement Learning (RL) to control the UAV. Aiming at the frequent undesirable setting of the traditional PID controller, this paper adopts the method of the reinforcement learning algorithm to set the PID parameters. Reinforcement learning is a well-known technique in the domain of Machine Learning (ML), which interacts with the environment and learns the knowledge without the requirement of massive priori training samples. Thus, the adaptive RL-PID algorithm has good adaptability and the parameter setting to support UAV flight tasks in unknown environments.