Recruiting Ant Colony System

We briefly describe an approach to extend the Ant Colony System by a group recruitment strategy. The underlying idea is very similar to the behavior of real ants, which use e.g. tandem-techniques to guide each other to large food sources. General investigations with small and medium sized colonies have shown that such strategies in combination with mass recruitment (pheromones) increase the food-supply rate. Surprisingly, this quite natural mechanism can be easily adapted for Ant Algorithms and has inspired our idea of the Recruiting Ant Colony System. The experiments with the optimization of a Traveling Salesman Problem instance show that RACS improves the auto-catalytic process. It seems that our new algorithm converge faster and more stable than the original one.