Real Time Bicycle Robot Balance Control Using Machine Learning

Abstract

Classic control systems leave the place to artificial intelligence control systems. In this study, an exemplary system is realized. In this study, the balance control of a two-wheel bicycle system was carried out. The control process is first was realized by using the PID control algorithm. Angle, acceleration and motor control signals in this process were taken as training data and a training model was created. As a result of the training, the code produced for the machine learning algorithm was embedded on the control card. The control of new balance conditions was carried out in real-time with machine learning.