Abstract & Overview
In this paper, we examine the theoretical cost function equivalence between Model Predictive Control (MPC) and Linear-Quadratic Gaussian (LQG) control, as well as Linear-Quadratic Regulator (LQR) control under specific conditions. Specifically, we linearize the Rotary Inverted Pendulum (RIP) system and construct a Kalman filter state estimator for application in both the LQG and MPC controllers with input and output constraints. We also assume measurable and computable states when designing the LQR controller. Through simulation and experimentation, we demonstrate that, despite the equivalence in cost functions, the output response of MPC is significantly better than that of both LQG and LQR. Our findings not only substantially bridge gaps in control theory but also emphasize the robustness of MPC in complex real-world applications. These insights pave the way for more effective and reliable control strategies across various engineering fields.