Ant Harvesting Simulation in Second Life
Andrew Cantino
April, 2007
Introduction

In this project, I have created a simulation of ant harvesting behavior. While I make no claims that the mechanics of this simulation are what real ants actually do, I was inspired by research into real ants, and so some parts of the simulation are arguably realistic.

My simulation is run in Second Life, an online 3D environment where all content is user created -- it is much like a 3D World Wide Web. As such, it is very appealing as an environment for testing and prototyping interesting ideas, both programmatically and geometrically. There are a number of science and math related projects in Second Life, including some involving Artificial Life [1] [2].

Real Ants

Army ants move and forage in large columns made up of thousands of ants. When an army ant finds some prey, it will head back to the column, leaving a chemical recruitment trail. When it encounters other army ants it will signal them to "search." It will continue to wander around the column recruiting ants for some time. When searching ants encounter the recruitment trail, they follow it to the prey. Some of them will also become recruiters. If no recruitment trail is found after some time, the recruited ants will continue on their way. [3]

Recruitment strategies vary from one ant species to another. A simple recruitment system seen in some ants (such as Camponotus sericeus) is dubbed "tandem running" [3]. In this strategy, a recruited ant must be lead to the prey through physical contact with a recruiting ant. In a more advanced form, called "group recruitment," (seen in C. socius) the ants are oriented by a recruiter and then follow a chemical trail. In "mass recruitment," ants will follow the trail after a recruiter tells them to "search" for it. This is seen in Solenopsis invicta and Eciton hamatum, and is what I model. Army ants hunt prey that can defend themselves against small numbers of ants, so this rapid recruitment is essential for successful harvesting behavior. [3]

My Simulation

Play the movie below to learn about my simulation, and to watch it in action. Please be sure to have your sound enabled.


Or, view this movie on YouTube.
Future Work

Further work could look at self-organization in ant colonies. For example, some ant species tend to organize such that each colony has its own area for foraging [4]. Additionally, further work could explore ant evolution. Ants are selected at the group level, not the individual level, which leads to colony-oriented behavior [5]. Finally, ants have many fascinating behaviors that would be worth modeling, including ant food type specialization, group hunting, colony movement, visual orientation, raids on other social insects [4], etc.

Press

I released a version of the above video on YouTube in early April, and it was picked up by New World Notes, a Second Life blog. The blogger, Hamlet Au, interviewed me and saw the ants in Second Life for the article. Subsequently, his article was picked up by Boing Boing, Kotaku, Grey Thumb, PacificRim Exchange, and Converjed.

Questions or comments?


References
[1] "Virtual Critters Thrive, Evolve in Online World," Mason Inman, National Geographic News, March 8, 2007
[2] "God Game," Wagner James Au, New World Notes
[3] "Mass Recruitment by Army Ants," Ruth Chadab and Carl W. Rettenmeyer, Science, New Series, Vol. 188, No. 4193. (Jun. 13, 1975), pp. 1124-1125.
[4] "Foraging Pattern, Colony Distribution, and Foraging Range of the Florida Harvester Ant, Pogonomyrmex Badius," Janet S. Harrison and John B. Gentry, Ecology, Vol. 62, No. 6. (Dec., 1981), pp. 1467-1473.
[5] "Ecology of Foraging by Ants," C. R. Carroll and D. H. Janzen, Annual Review of Ecology and Systematics, Vol. 4. (1973), pp. 231-257.