In this book we present a vision system for autonomous navigation of a mobile platform. The system is enable to interact with its immediate surroundings, recognizing obstacles and other moving objects, and obtaining a stable view of the world. In fact, a vision system for autonomous navigation has to detect the objects in the environment and classify them as “obstacles”, in order to avoid them, or as “target” objects, in order to follow them. We face the challenging problem of the “Obstacle detection and avoidance”. It consists of obstacle detection in order to find the safety path to follow during the autonomous navigation of a mobile platform. The major contribution of this work concerns a “perceptive” representation of the environment, that it is not a “passive” representation, but related to the final goal of autonomous navigation. It is based on the stereo vision paradigm and detect obstacles and moving objects in the scene right according to the autonomous navigation goal, that is obtaining a result as fine as it is enough for our aims. The results of our method for stereo vision are shown in a comparison with the best algorithms in the literature.