Reinforcement Learning Compensation based PD Control for Inverted Pendulum

In this paper, we present a Control Algorithm based on Reinforcement Learning for an inverted pendulum. By implementing the Q-Learning techniques in the PD control scheme, the pendulum is enabled to improve its online performance and adapt to changes in the parameters of the pendulum model. In a first step, Q-Learning is used so that the control can balance the pendulum towards its inverted vertical position; In a second step, we combine hybrid techniques of Q-Learning and PD control. With this combination, we can bring the pendulum to its inverted vertical position, regardless of the applied disturbance. Finally, the results of the simulation show the effectiveness of the proposed controller.