Neuro-fuzzy control system for vision-based autonomous vehicle

This paper describes a neuro-fuzzy control system for an autonomous vehicle to follow roadways with vision camera. Without a complex geometric reasoning from visual image to vehicle-centered representation as in conventional studies, the control system presented transfers the inputs of visual information into the output of steering angle directly. The neuro-fuzzy controller replaces the human driving skill of nonlinear relation between the points of road lane on the camera image and the speed and steering angle of vehicle on the real ground. In straight and curved road, the driving performances by the proposed control scheme are measured in simulation and experimental test.