Hybrid Scheme of Localization and Map Building with a LRF and a LPS for a Mobile Robot Navigation

Abstract

Localization and map building in an indoor environment are important problems for a mobile robot. There are two approaches to these problems: a metric approach and a topological approach. The metric approach uses various range sensors to determine the distance between an obstacle and the robot for the precise localization of a robot; the topological approach gives the robot its state on the basis of landmarks. Because both approaches have some advantages and disadvantages, researchers in the field of mobile robotics have combined the two approaches so that the robot can explore the environment autonomously. This so called hybrid approach still requires certain types of additional odometric information for rapid convergence of positioning the robot as well as information on landmark recognition. To build a more accurate map and to localize precise the position and orientation of the robot, we use a local positioning sensor to compensate for the accumulation of wheel encoder errors and landmark recognition errors. In this paper, we propose a new way of fusing a laser range finder, a local positioning sensor, and wheel encoders. The experimental results confirm that the robot can construct its own environment and localize itself accurately and consistently.