Comparing Perfect and Default Hierarchal Rules in solving Animate Problem

Zaghlul Bashir, Lubna

	Comparing Perfect and Default Hierarchal Rules in solving Animate Problem
Al-Ma'mon College Journal
Article 1, Volume 0, Issue 22, 2018, Pages 298-326
Author
Lubna Zaghlul Bashir
Abstract
The animate problem is the problem faced by an artificial animal that has to learn how to survive in its environment. Behaviors are specified through differential equations, forming a global system made of behavior subsystems, which interact in a number of ways. In this work, subsystems interact together by using learning classifier systems. , to Illustrates the approach with the example of Change Chameleon Color (CCC). The CCC System is built of two-classifier subsystems working together, each classifier system teaches a simple behavior, first classifier system, simulated robot chase behavior i.e. teach robot to move single step toward moving Light, second classifier system, teaches the simulated robot mimetic behavior i.e. teaches robot to change its color according to the background color. The system as a whole has as its learning goal the coordinate of behaviors. The results show identifying and coding appropriate activation schemes are decisive for the performance of a control system and Learning Classifier Systems is a feasible tool to build robust simulated robot control system. A Simple experiment was executed for CCC Compare between using perfect and default hierarchical rules, and the results show that the set of rules working together, as a default hierarchy should perform as well as the perfect set of rules. Default hierarchies’ rules, containing fewer rules, than non-overlapping rule sets for the same problem. Default hierarchical rules also enlarge the set of solutions space with no increase in the size of the problem space.
Keywords
Animate; Robotics behavior; Default Hierarchies; Learning Classifier Systems

Statistics Article View: 70 PDF Download: 29