Design Of Intelligent Vehicle Path Following System Based On Image Recognition

Intelligent vehicle driving system includes several sub system such as lane detection, obstacle detection, path planning, path following and so on. Path following is a key part in the research of intelligent vehicle, which is widely concerned by domestic and foreign scholars in recent years. This paper proposes a path following control algorithm based on image recognition. And a 20:1 intelligent vehicle model with STM32F103 chip as the core of control module, was designed to accomplish the path following research and its control method research. Based on experiment, the path following algorithm and its control effect were validated.The main contents of this paper are as follows:(1) Designing the intelligent car system and building the vehicle model, according to the requirement of intelligent vehicle path identification and path following. The vehicle model controlled by a STM32F103 chip board is the hardware part of the intelligent vehicle system. It is formed by camera module, Bluetooth serial communication module, power supply module, motor drive module, velocity measure modules and circuit connection between the modules.(2) The path following algorithm for intelligent vehicle system. In order to make the vehicle follow the path efficiently and stably, this paper divide the angle of path deviating from right ahead into several sets. Each set corresponds to a running mode. And this is so-called qualitative input, learning from fuzzy control algorithm. The path following control is realized by the combination of coarse and fine adjustment of the direction of the running vehicle.(3) The velocity measure method and the PID control method of the intelligent vehicle motors. The vehicle model is a three-wheel model, comprising two driving wheels and an universal wheel. Two driving wheels rotate at different speed to make the vehicle turn direction。The PID closed-loop control method is used to control the motors, and the measuring period method, which is widely used for the measurement of low speed, is applied to the measurement of the wheels.(4) Through the simulation experiment, this paper makes a comparison between the algorithm and the traditional preview control algorithm which can get the path information by the least square method, and compares the computation and the accuracy of the two algorithms. The result of the experiment doing by the model of the vehicle shows that the vehicle can be driven by the planning path. This proves that the embedded platform based on STM32F103 board, and the path following algorithm based on image recognition is correct and feasible.