Ant coverage using diffusion at and between grids

Abstract

This research looks into how an ant robot using Node Counting without diffusion compares to one using Node Counting with diffusion at repeatedly covering the bottleneck sub-grid. This is done by running a simulation in which the agent, the one ant robot used, covers the bottleneck sub-grid an increasing number of times using both algorithms. The bottleneck sub-grid is a three armed square that connects three sub-grids together. Efficient coverage on this grid is important in order to efficiently cover grids that contains the bottleneck sub-grid in general. When executing the first run on a not yet covered grid the Node Counting algorithm without diffusion performs with a success rate of 0.83, better than the Node Counting algorithm with diffusion that achieves a success rate of 0.75. When repeatedly covering the bottleneck sub-grid, the success rate of Node Counting without diffusion converges to 0.71 and that of Node Counting with diffusion to 0.34, so diffusion has a negative effect in this case.