Two-wheeled Car System Design And Implementation Based On Vision Autonomous Navigation

Two-wheeled Car is the current hot field of robotics research. It has two wheels which is on the same axis. Two-wheeled car is similar balance pendulum which is a nonlinear naturally unstable system. Using acceleration sensor and gyro sensor to get two-wheeled Car’s body tilt information, then the controller will process the information and converted it into voltage of the motor to make the car balance. In order to further improve the level of self-balancing car intelligence, Combined with machine vision research results to design a self-balancing car which with autonomous navigation has important significance.The design of two-wheeled car system which with autonomous navigation has two key technical difficulties: the precise control of two-wheeled car and autonomous navigation. In this paper, Kalman filtering be used to reduce accelerometer’s high-frequency interference and gyroscope’s low frequency error. PID control algorithm be used to improve motor response speed and control accuracy to ensure good control results. In the realization of autonomous navigation, methods of denoising, binarization, image extraction be used to process image information in order to obtain the right path.In this paper, using MC9S12XS128 as the controller, combined with a gyroscope ENC-03, three-axis accelerometer MMA7260, camera OV7620, to design a small two-wheeled self-balancing car which can run stably on a specific road with autonomous navigation. Finally, the theoretical analysis and experimental tests show the feasibility and rationality of this design approach.