Extraction of Roads From Out Door Images

A few separate initiatives from the robotics community to design and prove a mechanical automated solution have taken place. Here at the Pontificia Universidad Javeriana, in Colombia, we are working in this problem: on a previous project the mobile robot Ursula was developed (Rizo et al., 2003); and now we are working in a new mobile robot called Amaranta (Figure1). One part of the humanitarian demining problem is the navigation, in the two projects the autonomous navigation task is executed based on a vision system that uses a camera mounted on the robot. Landmines could be placed in any type of terrain: deserts, mountains, swamps, roads, forests, etc. This means that when trying to build a robot for demining operations, its workspace has to be previously defined and limited. In this work humanitarian demining, is limited to operate on places that have been modified by man and that represent great importance for a community, for example the roads or paths. Specifically, the systems have been designed for the Colombian territory (Rizo et al., 2003).