Comparisons in Evolution and Engineering: The Collective Intelligence of Sorting

Abstract

Collaboration between biologists and roboticists can facilitate the creation of new behavioral algorithms by roboticists and help biologists by exposing the underlying mechanisms that allow the algorithms to function (for a review see Webb, 2000). This paper makes a direct comparison between robot annular puck sorting using real robots (Wilson, Melhuish, Sendova-Franks & Scholes, 2004) and brood sorting in the ant Leptothorax albipennis (Franks & Sendova-Franks, 1992). We compared the ants’ and robots’ structures in terms of radial displacement, shape, compactness, completeness and separation. This showed the effectiveness and limitations of using metrics developed for robots to quantify structures built by ants and helped relate common aspects of the structures to possible common aspects of the underlying algorithms. The ant behavioral rule set is still under investigation and a better understanding of the structure it creates has proved a very useful tool to examine aspects of the algorithm. We draw the conclusions that firstly, the size of the area available for sorting affects the robot algorithm and ant algorithm in opposite ways, secondly, the metrics have proved very useful in aiding discrimination between the robot and ant structures and thirdly, application of the metrics to the ant structure has offered biologists new ideas and a better understanding of the structure and how it is built.