Evolutionary robot action development based upon Intelligent Composite Action Control

Abstract

This paper presents a framework for evolutionary robot action developement to autonomously plan complex/diverse cooperative actions and skillfully perform them. Complex actions are realized based upon the Intelligent Composite Action Control, which is a learning methodology for intelligent robots that gradually realize complex actions from fundamental motions. The Multi-stage Genetic Algorithm (MGA) is used for efficient construction of action intelligence. And for autonomous planning of diverse cooperation according to the situation, Variable-chromosome-length Genetic Algorithm is introduced and combined to MGA. With demonstrative examples of cooperative robot soccer actions the process of efficient construction of the action intelligence is presented.