Real-time HIL Implementation of DDPG based Reinforcement Learning Controller for a DC servo motor with Inertia Disc and Rotary Inverted Pendulum

This paper presents a real-time Hardware-in-the-Loop (HIL) implementation of a Deep Deterministic Policy Gradient (DDPG) based reinforcement learning (RL) controller for the stabilization and trajectory tracking of a DC servo motor system coupled with an inertia disc and rotary inverted pendulum. The nonlinear and unstable dynamics of the system pose significant challenges to conventional control methods such as PID, which are often limited in adaptability and robustness. In this work, a model-free DDPG algorithm is trained in simulation and deployed in a real-time HIL environment to validate its performance under practical conditions. The controller learns continuous-valued control policies to maintain desired inertia disc position and balance the pendulum upright while managing the position of the rotary arm. Experimental results demonstrate that the DDPG-based controller achieves faster settling time, reduced overshoot, and improved stability compared to classical PD control, making it highly suitable for complex, safety-critical applications. This study highlights the potential of deep RL for real-time control of nonlinear electromechanical systems.