Reinforcement Learning based Parameter Optimization of Active Disturbance Rejection Control for Autonomous Underwater Vehicle

This paper proposes a liner active disturbance rejection control (LADRC) method based on the Q-Learning algorithm of reinforcement learning (RL) to six-degree-of-freedom motion an autonomous underwater vehicle (AUV). The number controllers is increased realize AUV decoupling. At same time, in order avoid oversize algorithm, combined with controlled content, simplified Q-learning constructed parameter adaptation LADRC controller. Finally, through simulation experiment controller fixed parameters and rationality effectiveness adaptation, unique advantages are verified.