Localization of a simple robot with low computational-power using a single short range sensor

Abstract

Localization is a fundamental problem in mobile robotics. Several methods have been proposed over the years, among them Monte Carlo Localization. This method is preferred among many researchers because of its robustness and ability to work with a variety of sensor types. Most localization implementations use equipment that provide high resolution, accurate and long range measurements. Such sensors however, are expensive and demanding in terms of power, computation and physical space. In this paper we investigate localization in the context of limited sensing. Our motivation arises from the fact that many applications require robots with limited resources. We examine and discuss the limitations of the single sensor setup and we test our proposed solutions experimentally.