Modeling and Control of Two Wheels Robot Using Linear Quadratic Regulators

Abstract

Robots play high role in industrial and healthcare systems. Two-wheeles self-balancing robots are based on inverted pendulum systems. In this paper, a two-wheeled self-balancing robot is built, modeled, and controlled. The robot uses an Arduino Uno board communicates with an Inertial Measurements Unit to get gyroscope and linear acceleration measurements. The tilt angle of the robot is calculated by fusing the angular velocity and the acceleration readings. The robot moves by controlling two DC motors with gearboxes using PWM signals. To use a Linear Quadratic Regulator, a linear model should be used to calculate its gain matrix. A second order linear model is obtained with the use of MATLAB's identification toolbox. The robot was initially controlled by using a PID controller only to obtain data for the identification process. The PID is tuned using trial and error method. The data used for identification were then obtained wirelessly using a Bluetooth module. The linear model showed a good prediction response when tested using another set of data. Finally, a number of Linear Quadratic Regulator gains were computed and applied to the real system. The computed Linear Quadratic Regulator gains showed different responses and were categorized into three cases. In which one category showed a very good response, one showed jittery response and one performed badly.