Application of Integrated Central and Distributed Decisions to Path Tracking Control of a Vision-based Intelligent Vehicle

Vision-based intelligent vehicle is widely used for its advantages in acquiring more information, higher flexibility in path setting, etc. The control architecture, which is the foundation of the intelligent vehicle system, must be appropriately built up before the vehicle system is designed. By analyzing the characteristics of the existing architectures, an architecture based on central and distributed decisions is presented. A specific vehicle system is designed based on path tracking. The vehicle system acquires the path information by machine vision and automatically tracks the recognized path. In order to recognize the path image robustly, the methods of segmental optimal thresholds and adaptable thresholds are presented. Furthermore, a preview controller, which can adjust the vehicle speed and its preview distance, is designed for improving the path-tracking accuracy. Finally, experiments are performed under different scenarios to verify the proposed theories and methods.