Ant Colony Simulation: Agent Based Models

Описание: I created a fully customizable agent based model to simulate Ant Colonies. The ants are the agents, and follow simple rules. They randomly search for food, and upon finding it they return to the hive, dropping a food pheromone which marks their trail. If another ant of the same colony is wandering and sniff a pheromone trail, they will likely follow it. The simulation supports multiple colonies, and different ant species can be created with different properties and stats. If two ants of a different hive encounter each other they will fight, dropping a war pheromone in the process. If any friendly ants smell that pheromone they will help in the battle. Individual ants can die, and have health. The hive is responsible for "feeding" it's ants at a variable rate. An ant will stay alive so long as it is not killed, and is continually fed by the hive. If the hive runs out of food stockpiles, it's ants will die and it cannot produce new children.

You will see two species of ants in this video, the blue species drops blue food pheromones, and the green drops green food pheromones. Both species drop pink war pheromones. I can interact with the simulation by placing food in the world.

From this simple model surprisingly complex behavior emerges. Ants will exploit the food sources closest to their hive first. They will create shortest paths between the source and their hive. It is not shown in this video, but ants will "spawn camp" another hive, surrounding the hive and killing the babies that emerge. You will also notice that the ants tend to carve out territories for their hive.

This project was written in c#, you can find more info and the source code here:
http://jdevh.com/2012/05/20/agent-bas...