Design And Control Research Of Self-balancing Two-wheeled Mobile Robot

Self-balancing two-wheeled mobile vehicle, also known as two-wheeled self-balancingrobot, is one of the branches of wheeled robots. It is ideal for limited space and occasionrequired large steering angle because of its compact structure, small size, flexible mobility.And it has great practical value among the military, civilian transportation, factory automationand other fields. Meanwhile, the self-balancing two-wheeled vehicle is statically instable,dynamically stable system, with nonlinear, strong coupling, multi-variable, under-actuatedcharacteristics. Therefore, it is a classic platform to test and validate a variety of controlmethods and has important theoretical research value.The issues of control research of self-balancing two-wheeled mobile robot are realizingits dynamic balance, servo tracking of expected motion, fully autonomous motion and so on.The balance control is premise and vital important. The research status in and abroad wasintroduced in this dissertation. Based on these researches, a robot was designed andconstructed. And it was worked as a test platform to study related control methods. First, thekinematics and dynamics analysis of the robot were proposed based on the principles ofNewtonian mechanics. Mathematical models of DC motors, wheels, robot body wereestablished, and according to the internal relations of these three components, themathematical model of entire system was established for further research. Based on thecharacteristics of the MEMS sensors used for measuring the posture of the robot, a sensordata processing and fusion algorithm was proposed based on Kalman filtering principle toprocess gyroscope and accelerometer data, so as to obtain the optimal estimation of bodyinclined angle and inclined angular velocity of the robot. The system was decoupled intolinear motion and steering movement two subsystems. And at the position of the equilibriumpoint, the linear model of linear motion subsystem was gained. Traditional PID controller,pole placement control algorithms were studied based on this simplified model respectively.Besides, a new learning based control theory was researched and a controller based onSupport Vector Regression was designed to achieve the dynamic balance.MATLAB was used for the controllers simulation. The built platform was used for datafusion and balance control experiments. The related results tested and validated the datafusion algorithm’s correctness and the corresponding controllers’ effectiveness.