EFFICIENT ROBOT SELF-LOCALIZATION IN SIMPLE POLYGONS

Abstract

Introduction One of the basic tasks faced by an autonomous mobile robot is the problem of self-localization, that is determining its position in its environment; this is also sometimes called the "where am I?" problem. There are many variations of this problem depending on the environment and the sensor data and a-priori information available to the robot. Here, we consider a robot that is given a map of its environment but has no knowledge of its location on the map. Often, it is assumed that this problem can be solved by only using sensor data and allowing the robot to make a small local pertubation of its current position. But if there are self-similar parts in the environment, then this approach may not suffice to distinguish between the possible locations of the robot. This issue has been addressed in numerous contexts, ranging from aerial photography to autonomous vehicles for the exploration of landscapes. We consider an idealized version of the problem