Reinforcement Learning and Sliding Mode Hybrid Controller for an Enhanced Inverted Pendulum on Cart System | ||
| مهندسی مکانیک مدرس | ||
| Articles in Press, Accepted Manuscript, Available Online from 06 December 2025 | ||
| Document Type: مقاله پژوهشی | ||
| DOI: 10.48311/mme.2025.117163.82870 | ||
| Authors | ||
| Ali Jalali; Navid Mohammadi* ; Morteza Tayefi | ||
| Institute of Intelligent Control Systems, K. N. Toosi University of Technology, Tehran, Iran | ||
| Abstract | ||
| This paper presents a comprehensive control study of an inverted pendulum on cart system enhanced with torsional spring-damper dynamics and cart damping. A Linear Quadratic Regulator (LQR) was designed using a linearized model, while Model Predictive Control (MPC) and Reinforcement Learning (RL) controllers were augmented with Monte Carlo simulations to evaluate robustness and sensitivity. Results demonstrated comparable performance among all methods in linear regimes, with stabilization times under 20 seconds and overshoot variation of 3%. To address nonlinear dynamics, a hybrid SMC-RL strategy was developed, reducing settling time and improving capability of maintaining stability under nonlinear behavior and large initial angles to 120-150°. The proposed SMC-RL framework achieved a success rate in stabilizing the system from diverse initial conditions, significantly outperforming standalone controllers in transient response and adaptability. System stability was formally verified through Lyapunov analysis and empirically confirmed by Monte Carlo simulations, which demonstrated consistent performance with minimal standard deviation across 80 randomized trials. | ||
| Keywords | ||
| Hybrid SMC-RL Controller; Reinforcement Learning; Model Predictive Control; Monte Carlo Simulation; Inverted Pendulum on cart | ||
|
Statistics Article View: 181 |
||
| Number of Journals | 45 |
| Number of Issues | 2,171 |
| Number of Articles | 24,674 |
| Article View | 24,447,984 |
| PDF Download | 17,555,117 |