Mimesis "Teori, taklit-benzetme teorisi dir.
Bu teoriden yola çıkılarak , robot beyinlerde hareketlerin, duyguların idrakini tesis edebilmek amacıyla sezgi ve algılamaların sinirlere yansıtılarak sembollere dönüştürülmesiyle davranışların doğru tepkilere dönüşerek ortaya çıkarılması üzerine kurulmuş ve bu süreçte doğru odaklanmayı sağlayacak sistemin oluşturulmasıdır..

Bu aşamalar için:
1) sembollerin ortaya çıkması
2 ) davranış tiplerini tanıması,
3 ) davranış tiplerine göre kendi sembollerini oluşturması
4 ) hareket primitiflerini elde etmek.
Sonunda ise bu metodun yapılabilirliği, bir humanoid robotunda farklı bir çok denemeler sonucunda ancak kabul görür.
Abstract
“Mimesis" theory focused in the cognitive science eld and \mirror neurons" found in biology eld shows that the behavior generation process isn't independent of the behavior cognition process. The generation and cognition processes have a close relationship each other. During the behavioral imitation period, a human being doesn't practice simple joint coordinate transformation, but will acknowledge the parents behavior. It understands the behavior after abstraction
as symbols, and will generate one's self behavior. Focusing on these facts, we propose a new method which carries out the behavior cognition and behavior generation processes at the same time.
We also propose a mathematical model based on Hidden Markov Models
in order to integrate four abilities
1) symbol emergence,
2) behavior recognition,
3) self behavior generation, and
4) acquiring the motion primitives.
Finally, the feasibility of this method is shown through several
experiments on a humanoid robot

Introduction
The research of humanoid robots has a long history and has accumulated a substantial amount of literature. The focus of early eorts was mostly on the dynamics and control of bipedial walking motion. Although it has not yet reached the level of a complete solution, with liability and adaptability, the hardware technology has been established for building autonomous humanoids
Recently, the human behavioral science and intelli gence has become conspicuous as a real research issue of robotics. Although the motivation of the articial intelligence originated there, the physical limitations have orcedorjustiedtheresearcherstocarryontheir research in a limited scope of complexity. It would be a major challenge of contemporary robotics to study robotic behaviors and intelligence in the full scale of complexity. This could then mutually share research outcomes and hypotheses with the human behavioral science and human intelligence.
The discovery of mirror neurons4] has been a notable topic of brain science concerning such issues. Mirror neurons have been found in primates' and humans' brains, which re when the subject observes a specic behavior and also re when the subject starts to act in the same manner. It also locates on Broka's area, which has a close relationship with languagemanagement. Thisfact suggeststhat the behavior recognition process and behavior generation process are combined as the same information processing scheme. This scheme is nothing but a core engine of the symbol manipulation ability. Indeed, in Donald's “Mimesis Theory", he said that symbol manipulation and communicative ability are founded upon behavior imitation, which is integration of behavior recognition and generation. We believe that a paradigm can be proposed taking advantage of the mirror neurons, with considerations of Deacon's contention that language and the brain had caused each other to evolve.



Sofar, manyresearchershavetackledwiththe issues between the imitation learning for humanoids and humanintelligence. There are some suggestions that module structure of basic motions is needed for the symbolizationand representationofcomplex behavior,
such as Schaal's wor. In Kuniyoshi's approach robots can eproduce complex behaviors from observation of human demonstration with the abstraction and symbolization, however, it is dicult to be applied to general recognition and reproduction process because of lack of dynamics point of view, that means therobotshavetomemorizethewhole owofbasicbehavior. Moreover, the basic behavior modules needed to be designed by developer. Samejima et al haveproposed an imitation learning framework with symbolization modules. In this case, a premise have been set that sequence of symbol is given from others by communication, thus a certain representation model for dynamics of the whole body motion are needed. In this paper, we propose a mathematical model that abstracts the whole body motions as symbols, generates motion patterns from the symbols, and distinguishes motion patterns based on the symbols. In other words, it is a functional realization of embodied symbol emergence framework which is an inspiration from the mirror neurons and the mimesis theory.
Therefore we call the framework as "mimesis model".
The purpose of the research is to propose a methodology of mathematical design for the mimesis mode



makalenin devamı
kaynak PDF