Deep Q-Learning Enhanced Iterative Learning Control


  • Mustafa Kutlu Sakarya University of Applied Sciences



Iterative Learning Control (ILC), Deep Q-Learning (DQL), Hybrid Controller


Iterative Learning Control (ILC) is a method used extensively in repetitive control tasks to improve performance over time by learning from past iterations. However, traditional ILC methods often struggle with dynamic environments and require manual tuning for optimal performance. This study introduces a novel approach by integrating Deep Q-Learning (DQL) with ILC, forming an enhanced system. The DQL component adaptively tunes the learning parameters of ILC based on performance feedback, aiming to improve error reduction, adaptability, and convergence speed. The methodology involved developing a custom simulation environment to test the system under various conditions. The system was evaluated based on its ability to reduce cumulative error, adapt to changes, and achieve faster convergence. The results demonstrated that the system significantly outperformed traditional ILC methods in all assessed metrics.

Author Biography

Mustafa Kutlu, Sakarya University of Applied Sciences

Mechatronics Engineering Department,  Turkey




How to Cite

Kutlu, M. (2023). Deep Q-Learning Enhanced Iterative Learning Control. AS-Proceedings, 1(7), 1033–1038.