Improving Cooperative Robot Exploration Using an Hexagonal World Representation

Abstract

In this paper, we used an alternative to the frequently used quadrangular grid map as representation of the world. This alternative uses an hexagonal grid instead a quadrangular grid to represent the world where simulated robots explore. Both representations used an occupancy grid map system, where the world is divided by many cells (hexagons or squares) and only one element can be in the cell at the same time. We made several experiments to observe how exploration efficiency is affected using both representations and four different exploration algorithms. Experiments were made in two scenarios. The first scenario tested the exploration efficiency of a single robot using different number of random obstacles. The second scenario tested the exploration efficiency using different number of cooperative robots leaving the number of random obstacles as a constant. We used SimERl software to make all exploration experiments in a simulation environment. Experiments results showed that hexagonal grid map representation was better than quadrangular grid map representation if the world is very big or if the exploration algorithm is not very efficient. Improvement in efficiency could be small if the exploration algorithm is very efficient, cooperation among exploration robots is very good, the area to explore is small or if the number of exploration robots is big. Nevertheless, even if improvement in efficiency is small, the variance in hexagonal representation is lesser. Reduction of variance in consecutive exploration experiments is an advantage because we are interested in explorations that always have a good degree of efficiency.