Evolution of intelligent agents:: A new approach to automatic plan design

Evolution of intelligent agents, a new approach to automatic planning, is presented here for the first time as a new technique for evolving systems capable of generating "optimal" plans without any prior knowledge of the environment, or any method (i.e. schemata) concerning plan design. The classical model of system-environment interaction is extended by making it more "natural". By the use of the recent gene expression programming (GEP) technique a fully functional, multi-layered system architecture capable of solving complex tasks is proposed. The power of the new approach is demonstrated by testing it on several micro-world problems. Copyright (C) 2003 IFAC.