Evolving Autonomous Robot: From Controller to Morphology

Abstract

Introduction Artificiallife recia h comple(6 ts traditional biological scie---W thatconce:C analyse of livingsyste(' as it synthe#'#C artificialsystei that canen(';# the be--- viors of natural living syste(# In thisres('fl h fie'# e olutionaryte hnique have be' wideC applie to e olve syste# thateat(# life#fl--- e be' viors. In re(' t ye:W' the artificiallife te hnique (i.ee e olutionary algorithms) has be# prop ose to synthe';# control systeS for robots [1], [2].The ce tral ide is that by usingthe characte#C#SC ofse:'W#(6#C#SC( of e olutionary algorithms, we can synthe### robot be' viors close andclose to what we e# e#C ove rthe geW': tions and throughthe fitneh improve'S t. Base onthe e olutionaryte hnique eiqu ye' work in e olving robot controlleC can be catefl(6#SW intothe Ge#W; Algorithm(GA)-base work(e('; [2]-- [4]) andGe(:WS Programming(GP)-base work(e('--- [5], [6]). IngeW':W( aGA-base work involve ev( dingthe re--- antparame6C' of a controlle (such asthe we---' ts and