Search

This paper presents the design of multi-robots or multiple mobile sensors and gives the results of the team behavior of them in a potential field in a collaborative manner. The aim is to prove the concept of the fractal conductivity-based approach to mobile sensor networks in a potential field. We attempted to create multi-robots having similar inherent nature, similar background knowledge, and the same mission as relevant bodies, the creation of some sort of attraction force between them is achieved with the application of fractal conductivity of belief functions. We attempted to develop multi-robots capable of carrying out an orchestrated movement in an environment having some kind of potential field such as a chemical substance contaminated into sea or lake water. The basis of the theory of behavior in a potential field is established. The fractal conductivity approach is explained. A mobile sensor structure having an input, output, and internal belief functions is designed. Internal structures of mobile sensors are implemented by software. A mobile sensor network composed of mobile sensor nodes is developed. Two communities of interest groups have been constituted. Collaborative behaviors of multi-robots in a potential field are investigated. The mobile sensors move towards a target by sensing the potential field at their locations, broadcasting the information composed of locally sensed intensity of the potential field around them and the infrastructure knowledge of themselves to other sensors and receiving the information of the potential field intensities together with the identity information that is being broadcasted from the other mobile sensors inside the potential field. The target is a source creating the potential field. A mobile sensor is a body incorporating a prior knowledge base that is the identity information in the form of a fractal belief function inside itself. After developing the theoretical basis for sensing potential field and movement, a fractal conductivity approach, which is based on fractal natures of the belief functions, is applied to this basic approach to obtain the movements of the multi-robots as a coordinated team, i.e., grouped as relevant bodies. The operation of mobile sensors based on time division multiple access method is achieved. The results of coordinated movement and obstacle avoidance have been demonstrated by the simulation results.

Bu çalışmada, web sitelerine erişimi etkin hale getirmek ve kullanıcı gereksinimlerini desteklemek amacıyla geliştirilen bir portal yazılımının sağladığı temel servisler ve geliştirme aşamaları özetlenmiştir. Geliştirilen portal yazılımının temel mimari yapısı, bu mimari yapı içinde sağlanan kişisel sayfa düzenleme, erişim yetki düzenlemesi, e-posta ve benzeri sistemlerle entegrasyon, proje yönetimi ve benzeri hususlar detaylı olarak açıklanmaktadır. Ayrıca, portal yazılımının geliştirilmesinde kullanılan teknolojiler ve yazılım araçları özetlenmektedir

öz SİNİRSEL DENETİMLİ FRAKTAL KARAR A?LAR VE ÇOK PARMAKLI ELLERİN KAVRAMA BİÇİMLERİNE UYGULAMASI NAZLIBİLEK, Sedat Doktora Tezi, Elektrik Elektronik Mühendisliği Ana Bilim Dalı Tez Yöneticisi : Doç. Dr. Aydan Müşerref ERKMEN Eylül, 1993, 269 sayfa Bu tezde sunulan çalışmanın amacı eksik sezilmiş ve işlenmiş giriş verisine dayalı bir karara göre bir eylem yapan bir sistem geliştirmektir. Sistem üç ana modülden oluşmaktadır, bunlar Sezme-Algılayıcı modülü destek fonksiyonları biçiminde görüntüye dayalı bir delil üretir. Sezme-Algılayıcmm girişleri robot çalışma ortamında bulunan farklı nesnelerin görüntüleridir. Buradaki söz konusu nesneler ortamda bulunan ve eksik bilinen robot elinin kavrayabileceği parçalardır. Nesneler eksik bilindiği için, bunların düşük seviye görüntü işlenmesinde, başlangıç destek fonksiyonlarını tahsis etmek üzere fraktal tabanlı bir yöntem geliştirilmiştir. Bu çalışmada, çok parmaklı robot elinin göreve dayalı ön el şekillenmesini belirlemek için yeni geliştirilmiş Fraktal Karar Ağı kullanılmaktadır.1 Sinirsel Denetleyici Sezme-Algıl ayıcı modülün girişine en ilgili bölgeleri belirleyerek Fraktal Karar Ağını yerel olarak etkinleştirmede kullanılır. Anahtar Kelimeler : Sınıflandırma, Fraktallar, Destek Fonksiyonları, Sinirsel Ağlar, Robotlar. Bilim Dalı Sayısal Kodu : 619.04.06 vi ABSTRACT NEURO CONTROLLED FRACTAL INFERENCE NETWORKS AND THEIR APPLICATIONS TO GRASPING WITH MULTIFINGERED ROBOT HANDS NAZLIBİLEK, Sedat PhD. in Electrical Engineering Supervisor: Assoc. Prof.Dr. Aydan Müşerref ERKMEN September, 1993, 269 pages The purpose of the work presented in this paper is to develop a system making an action according to a decision based on imperfectly sensed and processed input data. The system consists of three main modules, namely Sens-Perceptor (SP), Fractal Inference Network (FIN), and Neuro-Controller (NC). Sens-Perceptor is developed upon a new methodology for generating the perception input to a Neuro Controlled Fractal Inference Network. The Sens - Perceptor module generates image based sensory evidence in the form of belief functions. Inputs to the Sens-Perceptor are the images of different objects contained in the robot workspace. The objects considered here are imperfectly known graspable parts available to the robot hand in its environment. Since the objects are imperfectly known at different levels of iiiresolution a fractal based methodology is developed for assigning initial beliefs at the low level image processing of objects. In this study, the FIN is developed as an inference network where information propagation is based on fractal conductivity and is used for determining the task oriented hand preshapes of a multifingered robot hand. The Neuro Controller locally activates the FIN by determining regions of high relevance to the input of the Sens-Perceptor module and thus reduces activation leakage over the network due to propagation under low relevance. Keywords: Classification, Fractals, Belief Functions, Neural Networks, Robotics. Science Code: 619.04.06 IV 269

108