Obstacle avoidance and self-localization system for autonomous vehicles

Abstract

In this paper, a system for vision based autonomous navigation is presented which is capable of obstacle avoidance as well as self-localization. At first a ground plane detection method is proposed which is capable of working on different most commonly occurring indoor/outdoor ground textures. The benefit of ground plane detection is that the system is not dependent on some specific property of obstacle to detect (such as size and orientation). Next, a strategy for vehicle localization and implementation of proposed method is discussed.