Reinforcement Learning and Fuzzy PID Control for Ball-on-plate Systems

Ball-on-plate system (BPs) is a well-known two-dimensional balancing problem that many types of controllers were implemented. The Reinforcement Learning (RL) and Fuzzy PID controllers are proposed to improve the performance of time-domain response. The comparison of the controllers with PD controller shows the high robustness of disturbance rejection and the requirements of performance are satisfied. The application of RL becomes a solution with different characteristics from other traditional controllers, and the performance of Fuzzy PID is confirmed in the BPs control.