Computer Vision Technology for Vehicular Robot to Follow Guided Track Using Neuro-Fuzzy System

摘要

This paper describes computer vision technology for a camera-based automatic guided vehicular mobile robot using neuro-fuzzy control system to follow guide lanes with a camera. Without a complicated geometric computing from a camera image to robot-position-localization in conventional researches, the proposed control system transfers the inputs of image sensor into the output of steering values directly. The neuro-fuzzy controller replaces the human driving skill of nonlinear relation between vanishing lines of guide lanes on the camera image and the steering angle of vehicular robot. In straight and curved road, the driving performances by the proposed control scheme are measured in simulation and experimental test.