An Autonomous Vehicle Control Strategy to Imitate Human Behavior: Applied in Path Tracking

摘要

This paper presents a personification of autonomous vehicle control strategy to accomplish curved path tracking, including three levels: curve fitting, curve discretization, and designing tracking controller. Bezier curve fitting is used to fit the path generated by RRT to create a smooth path. The purpose of discretization is to discrete the curve into points, and the Douglas-Peucker algorithm widely used in GIS field is employed to discrete the curve into set of lines. Controller design is on the basis of abstractions of the human decision thinking; a P controller combined with a PD controller mimic human driving behavior. Simulation results show that: the strategy on the condition of low speed (1-10m/s) demonstrates good speed robustness and tracking accuracy.