Genetic Algorithm Based Stabilization of Rotary Inverted Pendulum by State Feedback

Abstract

A Rotary Inverted Pendulum is a standard test-bed in control theory due to its nonlinear and unstable characteristics. Driving the pendulum to an upright position, sustaining it motionless there, and carrying on signal tracking tasks at this position are the main control applications related to this test device. This paper presents a comparative study of four state feedback controller structures performing pole placement that theoretically stabilizes the pendulum at an upright position and in real-time application. The first state feedback controller is established based solely on the pendulum dynamics. In contrast, the other controllers are developed due to a multi-criteria optimization procedure by running a Genetic Algorithm search. The Quanser kit is used to obtain and implement the controllers and to test their performance in real-time.