This thesis attempts to solve food collection problems using genetic programming. The genetic program will evolve programs that mimic the way ants can collect food and bring it back to a nest. There are two special factors in this genetic program that make the ants work together in order to solve the problem, as opposed to each ant acting on its own. First, there will be a stream in the environment that the ants must cross to get to the food. Although all ants have the same program, some must move into the water and die, building a bridge for the other ants to cross. The surviving ants must realize that a bridge has already been built that they can use, instead of killing themselves by building another bridge. Second, the food will be too heavy for one ant to lift alone. The ants must find the food, and call to other ants for help. If all of the ants are at food waiting for help, some, but not all, of the ants must realize that they are in a deadlock situation, and leave their food to help other ants. Both of these problems require the ants to use teamwork to solve the problem. The ants must realize what other ants are doing, without direct communication or a state machine within the ants.
Library of Congress Subject Headings
Ants--Food--Computer simulation; Genetic programming (Computer Science); Genetic algorithms; Ants--Behavior--Computer simulation
Department, Program, or Center
Computer Science (GCCIS)
LaLena, Michael, "Teamwork in genetic programming" (1997). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus