A Study of Genetic Programming on Cooperative Model for an Ant Colony

• Main Authors: Yueh Hung Chen, 陳岳宏
• Other Authors: Koun Tem Sun
• Format: Others
• Language: zh-TW
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/89422331726795754741

碩士 === 臺南師範學院 === 資訊教育研究所 === 88 === A cooperative model is the action strategy in which every individual cooperates with others in a team. Thus, the better cooperative model the team has, the higher performance it has. However, every individual in the team can not detect the state of others. It only detects the change of the surroundings, and all individuals in the team are act independently. So, it is difficult to find an efficient cooperative model. The genetic programming (GP) can program automatically by simulating the evolutionary mechanism and process on computers. When the programs are evolved by using GP technique, an efficient program often can be generated for solving certain problem that we don’t understand in detail how to solve it. In fact, the performance of some evolved programs is better then that of the programs designed by experts. So, in this research, we try to use GP to evolve the cooperative model for the ant colony under both the single program architecture and the multi-agent architecture. Then, ants can cooperatively transport the food to the nest with these cooperative models. In our research, we use simple functions and terminals instead of complex ones to evolve the cooperative model for the ant colony. From the simulation, an efficient cooperative model can be evolved with both the single program architecture and the multi-agent architecture. So, it is proved that the GP technique can be used to evolve an efficient cooperative model. This research can not only can be used to related research, but also can be applied to develop more complex cooperative models (eg. the cooperative models for robots or cooperative learning model).