Title: Application of ant colony optimization to optimal foragaing theory: comparison of simulation and field results Authors
|Legaspi, Benjamin -|
|Isaacs, Jason -|
|Foo, Simin -|
Submitted to: Florida Agricultural and Mechnical University(FAMU) Research Journal
Publication Type: Proceedings
Publication Acceptance Date: January 11, 2010
Publication Date: March 15, 2010
Citation: Legaspi, B.C., Legaspi, J.C., Isaacs, J.I., Foo, S.Y. 2010. Application of ant colony optimization to optimal foragaing theory: comparison of simulation and field results. Proceedings of the Florida Agricultural and Mechnical University(FAMU) 1st Annual Research Summit. 9:134-154. Interpretive Summary: Ant Colony Optimization (ACO) refers to a family of computing methods, inspired by the behavior of real ants, and used to solve engineering problems. Real ants follow chemical trails secreted by other ants, and also excrete their own trails for other ants to follow. Individual ants simply follow the chemical markers and the colony as a whole tends to find the shortest path to a food source and back. ACO has practical applications ranging from package and vehicle routing, to telecommunications and the military. Scientists from USDA-Agriculture Research Service, Center for Medical, Veterinary and Agricultural Entomology, Gainesville, Florida, Florida A&M – Florida State University College of Engineering, ACO has been applied to the problem of random number generation. Evolutionary theory predicts that biological organisms will use food sources in specific patterns so as to optimize their time and energy. These predictions were compared against those using ACO computer simulations. In field experiments, red imported fire ant colonies were presented with different qualities of food sources at different distances from the colony. Evolutionary theory and ACO simulations were in general agreement. Future research may be helpful in investigating ant feeding behavior using ACO simulation. Possible applications may include data mining and scene analysis, as well as controlling Autonomous Robots/Vehicles.
Technical Abstract: Ant Colony Optimization (ACO) refers to the family of algorithms inspired by the behavior of real ants and used to solve combinatorial problems such as the Traveling Salesman Problem (TSP).Optimal Foraging Theory (OFT) is an evolutionary principle wherein foraging organisms or insect parasites seek to derive optimal use of time in consuming prey or parasitizing hosts. Here we employ computational techniques borrowed from ACO to study optimal foraging behavior, using the red imported fire ant (Solenopsis invicta) in field experiments. The ACO-OFT algorithm models the selection of paths leading to food resources of varying amounts and varying distances to a central colony. Distance, amounts of food and ant pheromone are determinants in selecting one path among multiple options. As such, ACO-OFT is both multiple objective (minimize distance, maximize food) and dynamic (food is diminishing). Simulations produced expected results, such as preferences for closer or higher quality food resources. Field data often did not conform to predictions of OFT, possibly because of uncontrolled environmental factors. The ACO-OFT algorithm may be improved in the path selection subroutine. For example, more realistic simulations may result from increasing the effect of pheromones, especially early in the run because of the reinforcing effect on path selection. We discuss possible engineering applications of ACO-OFT in the context of combinatorial problems including those in the TSP-family. We also discuss applications in data mining and scene analysis, as well as controlling Autonomous Robots/Vehicles (ARVs).