Integrated High-gain Feedback Control and Deep Reinforcement Learning Control of a Flexible Link With End-effector

Achieving precise trajectory tracking control of the end-effector of the flexible link is challenging due to the complexity of its dynamic equations. To address this challenge, this paper proposes a hybrid control method that combines a high-gain feedback control (HGFC) algorithm based on Lagrangian dynamics equations with a deep reinforcement learning algorithm based on Soft Actor-Critic (SAC) to form the HGFC-SAC framework. The HGFC-SAC method aims to track the reference trajectory of the end-effector accurately without perceiving its own flexible deformations while suppressing the vibration issues caused by its flexibility. Simulation results show that, compared with the HGFC and SAC methods, the HGFC-SAC method has better control performances in terms of trajectory tracking and vibration suppression.