Visual guidance of an autonomous vehicle in natural environments based on reflexive control

Autonomous navigation in outdoor environments is a complex task that until today is realized with high cost systems. This paper presents a system for autonomous navigation in weakly structured and outdoor environments like mountain ways. The main goals are to design a robust, cheap and fast system. The proposed solution constructs a simple scene description by region segmentation and analysis of the region variation in a sequence of color images. Then, reflexive planning calculates a locally optimal travel direction from the available information.