Localization of mobile robot using visual system

Abstract

One of the main problems in mobile robotics is localization. This article aims to increase accuracy of localization of mobile robot using visual system. Mobile robot is taking images of environment containing artificial markers. Design of the markers was also one of the tasks. Markers can be used indoors and are placed on ceiling, where there is low number of other disturbing elements. They are designed to be able to split each marker into two parts—one is changeless and the second is changing with each marker. This provides an easy segmentation of all markers and then distinguishes between them easily as well. Localization is based on changes of markers angle and position in acquired images. During research of suitable algorithms, several approaches were tested to gain optimal results of localization considering time consumption and accuracy. The last part of this work deals with implementation of visual system and odometry fusion. That provides even more precise localization of the mobile robot.