| Title |
Development of a Reinforcement Learning-Based Control Education Platform Using Python and LW-RCP |
| Authors |
이종범(Jongbeom Lee) ; 이태건(Taegun Lee) ; 주도윤(Doyoon Ju) ; 이영삼(Young Sam Lee) |
| DOI |
https://doi.org/10.5370/KIEE.2025.74.1.118 |
| Keywords |
Reinforcement learning; Pendubot; Sim-to-Real Learning; LW-RCP |
| Abstract |
This paper proposes a reinforcement learning-based control education platform utilizing Python and light-weight rapid control prototyping (LW-RCP). The platform employs the Sim-to-Real technique, in which neural networks are trained in a Python-based simulation environment and applied to real systems. The trained networks are converted into a format compatible with Matlab/Simulink. The lab-built LW-RCP is used to implement a real-time controller under the Simulink environment by incorporating the converted networks. The proposed platform allows students to easily apply reinforcement learning theory to real systems, contributing to the integration of reinforcement learning control into control curriculum. The effectiveness of the proposed platform is demonstrated by implementing a reinforcement learning controller for the pendubot system. The implemented controller performs the swing-up and transition control and exhibits strong disturbance rejection and recovery properties. |