Multicopter PID Attitude Controller Gain Auto-tuning through Reinforcement Learning Neural Networks

Abstract

Multicopters continue to gain their applications in the real world and their control problem has attracted a great number of studies. Among several control techniques, the proportional-integral-derivative control method appears to play important roles in seeking a simple and efficient controller for a complex system like a multicopter. However, the gain tuning process of a proportional-integral-derivative controller is still a time-consuming task and, therefore, finding an automatic gain tuning method which saves time and ensures satisfactory control performance has become one of the most important efforts that researchers all around the world are nowadays undertaking. In this paper, we present a proportional-integral-derivative controller gain auto-tuning method using the reinforcement learning neural networks. Software and hardware-in-the-loop simulations were carried out to demonstrate the effectiveness of the proposed method.