Military Strategies Based On Ants' Movements
Sometimes viable battlespace tactics come out of left field. In this case, a system that in all likelihood will end up as a military asset, started life in several commercial videogames.
Researcher Antonio Miguel Mora García, together with supervisors Juan Julián Merelo Guervós and Pedro Ángel Castillo Valdivieso, of the department of Computer Architecture and Technology of the University of Granada, worked on engineering a new system based out of the numerous ant colony optimisation algorithms used in many modern games.
All ant colony optimisation algorithms are essentially swarm intelligence methods intended to guide a large number of independent entities as one cohesive mass. Thus, they not only pathfind, but they attempt to manage resources effectively in the swarm, whilst they pathfind.
Now typically ant colony algorithms work from a goal-backwards philosophy. Any possible route can be taken, meandering all over the map. However, once a goal is found, the 'ant' heads back from the goal to the starting point, by the quickest possible route it can, and alerts all the others of its find. Successive weaves of ants use this path to return to the goal, with minor variation. Over time, the most optimum path is found.
To be useful to the military, a slightly different approach was necessary, one of knowing roughly where the goal is, and working out from this initial knowledge, the optimal way to reach it, avoiding obstacles such as ambush points. Still, the basic principles remain the same, they have just been 'tweaked' somewhat.
The scientists of the UGR have developed a mini-simulator not too dissimilar to an strategic overview such as is found in a commercial war-game, in order to define the battlefield settings. Where possible, aerial photographs are converted directly into such maps (see image) via swift machine vision processes. The goal is highlighted on the map, known enemy positions and potential choke points also illustrated, and the 'ants' representing the team, set loose. The swarm runs through countless possibilities until they have the tactics that result in the fewest friendly casualties, greatest security and speed, down pat. Then they return these results, showing what they did.
Part of the results of this research work has been presented in several conferences, both national and international, and published in journals such as "International Journal of Intelligent Systems".