Localization of Autonomous Robot in an Urban Area Based on SURF Feature Extraction of Images

Abstract

An autonomous robot is now an internationally discussed topic to ease the life of humans. Localization and movement are two rudimentary necessities of the autonomous robots before accomplishing any job. So, many researchers have proposed methods of localization using external tools like network connectivity, global positioning system (GPS), etc. However, if these tools are lost, either the movement will be paused, or the robot will be derailed from the actual mission. In these circumstances, the authors propose an approach to localize an autonomous robot in a specific area using the given set of images without external help. The image database has been prepared and kept in the internal memory of robot so that image matching can be done quickly. The localization method has been accomplished using three algorithms: (1) SURF, (2) ICP-BP, and (3) EMD. In the evaluation, SURF has been found better than ICP-BP and EMD in terms of accuracy and elapsed time. The authors believe that the proposed method will add value to other methods using some external tools even when those tools are unavailable.