Autonomous Navigation of an Intelligent Vehicle Using 1-D Optical Flow

Abstract

In this work, controlling a vehicle so that it does not bump into the wall of corridors in buildings is taken up as a problem of visual feedback control. The purpose of this study is to carry out such control using visual information in the real world within real time. We use 1-D optical flow calculated from intensity data obtained using a CCD linear sensor, as the visual information. Then a qualitative index which indicates the proximity of the moving vehicle to the left or right wall is extracted from the optical flow. By this qualitative approach, stable processing of visual recognition is achieved without explicit calculation of the distance. This method is based on general and qualitative knowledge of the environment, therefore there are fewer restraints of environmental conditions. Additionally, because no numerical models of the environment or vehicle are used for control, there is no need to consider the effect of parameter errors which was a serious problem until now.