서강대학교

초록/요약 도움말

Recently, there are many trials to understand biological phenomenon and imitate it for creating Artificial-Life. This paper purposed to suggest a self-reproducing system for generating Artificial-Life by evolutionary algorithm and show its graphical results. For generating complete agent, at first, a root unit is splitted into plenty of units through continuous unit-division. Second, splitted units are differentiated into one of structural unit, neural unit, joint unit. Third, neural units generate several neural-networks that each connection is bound by a synapse. This development process is occurred in each unit and it depends on concentration of gene products(protein) and environmental condition of the unit. Each agent, After development process, calculates its fitness. The fitness is a result of adding scores of structural efficiency, migration length and joint activity. DNA selection for evolution is decided by this value. The selected DNA would be evolved by reproduction, cross-over and mutation to generate new DNA for next generation. As a result, we built a system of generating Artificial-Life by biological mechanism and showed the graphical results. In the future, This system can be used as a tool for generating another Artificial-Life with more various life phenomenon. Moreover, the mechanism that evaluate specific ability and evolve it is also can be used to creating artificial robots which is suitable to handle given tasks automatically.