System Design and Navigation Control for Vision based CyberCar

The vision based cybercar named JLUIV-V is an electric automatic intelligent vehicle developed by Jilin University. This paper first describes the system structure, the design of the navigation line and the numerals. Then, the image segmentation algorithms for variable illuminations and the BP neural network algorithm classifier are introduced. Some image segmentation results of different images under variable illuminations are offered to show the performance of the image segmentation algorithms and the classifier. Last, the fuzzy variable structure controller for autonomous navigation is designed in detail and the experiments in Jilin University and in the culture center of Jilin Province are carried out. The experimental results show that the algorithms for variable illuminations, the classifier, and the controller have good performances in robust, reliable and accuracy.