Co-Author: Alireza Derakhshan (kianosh@mip.sdu.dk)

Main-Supervisor: Henrik Hautop Lund (hhl@mip.sdu.dk)
Co-Supervisor: Yves Demazeau (yves.demazeau@imag.fr)
Co-Supervisor: Thomas Klitbo (klitte@mip.sdu.dk)

Abstract: This Master thesis explores how to adapt the Body Games playground using Muli-Agent Systems based upon live human agents. The Body Games playground consists of small tiles were each has one FSR sensor and is able to communicate with its four neighbours. Based on an 6x6 environment this thesis states that it is possible to recognise a child's behaviour exhibited on the Body Games playground. The thesis proposes a design of an artificial neural network from an agent oriented approach (BDI agent) that is able to recognising a child's behaviour and classify it within 8 different behavioural categorise; thus, enabling the playground to adapt according to the classifications. Two frameworks have en implemented, one for application development on the Body Games playground and one for the artificial neural network. A play was developed, implemented and tested, based on the first framework, on a group of children to collected their behaviour. Afterwards the data was used by the artificial neural network to classify the different behaviours. The result was that the overall behaviour of the children can be recognised with a success of 96% based on 5 training subjects and 25 test subjects.

Download Thesis (26MB PDF)

Supervisor: Ivar Balslev (ivb@mip.sdu.dk)

Project description (in Danish): Projektet har til rådighed en stor mængde røntgenbilleder og en tilhørende database med oplysning om tændernes tilstand, hvad angår caries. I projektets første fase laves et program, der regulariserer billederne, så den betragtede tand og dens omgivelser afbildes i et normaliseret billede. I anden fase laves et træningsprogram, der enten optræner et neuralt netværk eller laver en database, der kan bruges for en statistisk baseret klassifikation. I tredje fase testes det neurale netværk/den statistiske klassifikation, idet billeder, der ikke indgik i træningen, forsøges diagnosticeret.

Synopsis: Biological systems, one and multi layer feed forward networks, perceptrons, back propagation of errors, algorithms for training, Hopfields network, autoassociative networks, statistical pattern recognition, probability estimation, applications in image analysis, hyphenation etc.

Synopsis: As tool for the study of dynamic agents LEGO MINDSTORMS robots will be used, and the students will be introduced to the robots and different ways of programming such robots. After a short summary about traditional agents and interfaces, the crucial aspects and techniques of new emergent technologies, as artificial life and self-adaptive robots, will be introduced. Robotics, input-output relations, adaptive-robotics, neural networks and others artificial life techniques will be examined. A first practical section will follow and the students will be requested to run a small project in order to familiarize with these techniques. Afterwards, the lessons will start facing problems related to human-machine interaction and related interfaces design. Psychological (perception, memory, attention, communication, education, etc.) and aesthetical (morphology design, behaviour design, etc.) aspects will be included. The last part of the course will focus on all the dynamical aspects of such a human-machine relationship (timing design, complexity increase, machine adaptation to users´ abilities, etc). The course will end up with a final project where the students will have to design and realize a dynamical, either virtual or real, system with a particular human oriented interaction. As a natural prosecution of the course user-test methods (and/or few others mass communication topics) will be shown and their knowledge tested on the former project results.

Synopsis (in Danish): Organisationsstruktur. Organisationsændringer. Målsætning og beslutningsprocesser. Strategi. Ledelsesformer – opgaver, holdninger, roller. Arbejdsmiljølovgivning. Arbejdspladsvurdering (APV). Økonomiske modeller og nøgletal. Regnskabsanalyse. Investeringer og kalkulationer. Erhvervsret - aftaleloven og købeloven.

Synopsis: To give participants theoretical and practical understanding of modern artificial intelligence approaches to robotics. The course will enable the participants to use these approaches in practise with mobile robots. • Neural network control.• Evolutionary robotics.• Behavior-based systems. • Biologically inspired robotics. • Artificial intelligence. • Simulation-reality transfer.

Synopsis: The course will present the following concepts and technologies for developing flexible and maintaniable code: Programming with Designpatterns, Application of object-oriented frameworks and Class libraries, Component-oriented programming, Aspect-Oriented Programming, Reflection and metal-level programming, Concurrent object-oriented programming. Futhermore the course will discuss various problems related to programming issues such as delegation vs. inheritance, safe exception handling, object re-entrance, deadlocking.

Synopsis: After a general introduction to Multi-Agent Systems, we present the basic notions of the domain, insisting on methodological issues of how such systems may be conceived and realised. We show how MAS methodology is different from more traditional methodologies and we identify the best problems and the particular domains in which MAS seems to be the most promising paradigm to adopt. We approach MAS starting by splitting them into Agents, Environments, Interactions, Organisations and Dynamics. An important part of the presentation deals with the different types of "Interactions". Starting from Game Theoretic Interactions, followed by the communicational aspects of MAS and Interaction Protocols, we introduce different notions about Coordination, Coherence, Task Allocation and Negotiation. A discussion about "Organisations" and "Dynamics" of MAS follows. Having settled the different parts that constitute a MAS, enriched by MAS Learning issues, we then deal with the programming tools (from languages for building MAS to integrative enviroments) before presenting Multi-Agent Orented Programming (from agentcentred approach to alternative approaches where interactions or organisations may be considered at first). Implementation experiences are shared. In the final part of the presentation, examples of real-world applications are presented in various domains (vision, GIS, NLP, robotics, simulaiton of complex natural systems, software engineering, telecommunications, town and country development, Internet applications). We conclude by presenting current MAS structures (organisations and events) and we report some recent major MAS events. Emphasis is given to concepts, results, and algorithms that have direct applications to computational multi-agent systems. As part of the course, several exercises are analysed and programmed using a standard multi-agent oriented programming tool.

Synopsis: Overview of the components of a compiler. Lexical analysis. Regular expressions. Programs for generating scanners. Top-down parsing for LL(1) grammars. Recursive descent. Bottom-up parsing for LR(1) grammars. Table driven parsers. Programs for generating parsers. Semantic analysis. Type checking. Different forms of scopes for variables, especially in block structured languages. Code generation, including the handling of procedures and different control structures. Optimization including liveness analysis and register allocation.

Synopsis: Robot co-ordinates, direct kinematics, inverse kinematics, work space analysis, redundancy of robots, robot dynamics, dynamical methods for trajectory planning.

Synopsis: Acquisition, representation, transformation, preprocessing and analysis of digital images. Image understanding, segmentation, morphology, pattern recognition, Fourier methods, and computer vision.

Synopsis: Classical mechanics (kinetics, dynamics, statics, work, energy and momentum), thermodynamics, transport phenomena (diffusion, heat conduction), elasticity, motions of fluids, radioactivity, wave motions, sound, light, geometrical optics.

Synopsis: Electric fields and potentials, magnetic fields and magnetic properties of matter. Maxwells equations and electromagnetic waves.

Synopsis: Numerical treatment of simple physical differential equations, programming in MATLAB.

Synopsis: The fundamental steps of program development: - classes, objects and their interplay - specification - program design - implementation. Examples of algorithms and data structures.

Synopsis: Algorithms: correctness and complexity analysis, greedy algorithms, divide and conquer, dynamic programming, sorting, graph algorithms, lower bounds. Data structures: abstract data types, trees, dictionaries, priority queues, disjoint sets.

Synopsis: Solution of linear and nonlinear (systems of) equations, approximation of functions, numerical integration and differentiation, solution of ordinary and partial differential equations.

Synopsis: Finite automatas, push-down automatas, Turing machines, regular languages, context free languages, grammars, decidability and computability.

Synopsis: Differentiation and integration of the standard functions, including hyperbolic and inverse trigonometric functions. Taylors approximation of degree n for a function of one variable, lHospitals rule. Solutions of first and second order linear ordinary differential equations and methods for nonlinear first order equations. Complex numbers, the complex exponential function and the complex quadratic equation. Infinite series and power series, radius of convergence, representation of functions by power series and applications to solving differential equations. Vectors and matrices, linear transformations of n-dimensional coordinate space. The Jacobi matrix for a vector-valued function of several variables, Taylors formula in two variables and the classification of stationary points. Line integrals of vector fields and the existence of potential functions. Multiple integrals, surface integrals and the flow of a vector field through a surface. The theorems of Green, Stokes and Gauss and applications to calculations of line, surface, and volume integrals.

Synopsis: The aim of teaching linear algebra is to enable the student to: 1)determine the dimension of a subspace of a vector space, find a basis and transform it to an orthonormal basis in the case where the given space has an inner product (Gram-Schmidt orthogonalization); 2)use the connection between linear mappings and matrices; 3)compute the determinant of a matrix; 4)find eigenvalues and eigenvectors for a given linear mapping or matrix; 5)find an orthonormal basis consisting of eigenvectors for a given Hermitian linear mapping on a finite dimensional Euclidean space (diagonalization of Hermitian matrices); 6)compute the projection of a given vector on a given subspace of Euclidean space, in particular compute approximations of continuous functions by trigonometric polynomials; 7)reduce a quadratic form and draw the graphs for systems of polynomial equations; 8)apply the simplex method for determining optimal solutions to problems in linear programming; 9)use the duality theory of linear programming.

Synopsis: • General introduction to mathematics on computers. • Introduction to the computer program Mathematica. • Programming techniques: How is a mathematical problem translated into a problem that can be solved on a computer? • Symbolic integration and differentiation. • Visualization of mathematics.
Introduction to numerical methods. • Introduction to image processing, the Fourier transformation, differential equations and the Laplace transformation.

Synopsis: External and internal mechanical forces, the formalism of D´alembert, Euler-Lagrange and Hamilton, the theory of rigid bodies, multivariable constrained optimisation, the calculus of variations

Synopsis: 2D scanning and image acquisition with a camera, including the perspective transform. Digital image representations. Image compression. Image transformations, including filtering, colour slicing, erosion and dilation. Elements of image analysis including segmentation, pattern recognition and Hough transformation.

Synopsis: Requirement specification, object-oriented analysis and design. Design methods and tools. Object-oriented programming languages, libraries and tools. Design patterns.

Synopsis: Advanced concepts and technologies in software engineering (distribution, concurrency, security, reliability, availability, scalability, performance), software architecture (frameworks, patterns, components), advanced programming in Java (distributed objects, threads and object persistence) and other topics.

Synopsis (in Danish): Digitalt grundkursus: Talsystemer og koder, Boolsk algebra, kombinatoriske netværk, Programmmerbar logik. Microcontroller hardware: CPU arkitektur og funktion, Hukommelseskredsløb, Parallel og seriel I/O, A/D og D/A konvertering, Timere og interrupt. Microcontroller programmering: Programstrukturering, Assembler programmering. Digitale periferienheder og interfacing: I/O enheder, Motorstyring. Analog interfacing: Feedback-teori (negativ), Operationsforstærkere, Differensforstærker, Dioder og transistorer, Målebro.

Synopsis (in Danish): Passive komponenter: Resistorer, kapacitorer og selvinduktioner. Strøm og spændingskilder: Thevenin- og Nortonækvivalenter. Kredsløbsanalyse: Ohms lov, Kirchoffs love, superpositionsprincippet. Knudepunkts- og maskemetode. Kompleks repræsentation af strøm, spænding og impedans Effektivværdi og effekt. Transient og steady state analyse. Laplacetransformation som analyseværktøj. Impedanstilpasning. Halvledere: Fænomenologisk beskrivelse af ladningstransport i rene og doterede halvledere, pn-overgangen. Aktive komponenter: Dioden, diodemodeller, diodeligningen og arbejdslinier. Transistoren, simpel fysisk model, arbejdslinier og transistor som switch. Integrerede forstærkere: Ideelle operations- og instrumenteringsforstærkere, introduktion til forstærkerfejl, feedback koblinger.

Synopsis (in Danish): Von Neumann arkitektur: Bussystem, CPU (kontrolenhed, ALU og instruktionssæt), hukommelsessystem, I/O-system, cache memory, selekteringskredsløb, exceptions og interrupts, eks. på periferienheder (DMA, timer, UART, interrupt controller, programmable peripheral interface). Parallelle arkitekturer: Pipelining: Instruktionspipelining, aritmetrisk pipelining, schedulering, datahazard, controlhazard. RISC kontra CISC arkitektur: Load/store filosofi, registerbehov, superscalar, superpipeline, pipelineudnyttelse, compilerkompleksitet. Multiprocessor- og multicomputersystemer: Klassifikation af parallelle arkitekturer (SISD, MISD, SIMD, MIMD), forbindelsesnetværk (n-dimensional, n-cube), koblingsgrad, busarbitrering, prioritering, cacheproblematik. Dataflow og systolisk array arkitekturer.

Synopsis (in Danish): Sensorer: Sensorprincipper (resistive, kapacitive, enkodere, …). Karakteristiske parametre for udvalgte sensortyper. Elektronik: Operations- og instrumenteringsforstærkere; - ideelle operationsforstærkere, eksempler på anvendelser. - bias, offset, drift, støj, PSRR, CMRR, fejlfaktor, slewrate og generel feed back teori. - kompensationsnetværk. Passive og aktive filtre. Transistortyper (BJT, FET): - funktionsbeskrivelse og opbygning. Halvlederkomponenter på storsignal niveau: - tab og idealiseringer. - eksempler på dioder og deres karakteristika - eksempler på anvendelser til switching. Simulering af kredsløb (Pspice). Elektriske aktuatorer: Beskrivelse af enkelte vigtige aktuatortyper (f.eks. DC og stepmotorer, moving coil devices.) ved hjælp af karakteristiske parametre og deres sammenhænge. Der lægges vægt på de parametre, der har betydning for interfacing på driversiden.

Synopsis (in Danish): Fagligt: - Gennemførelse af et projekt med udgangspunkt i et projektoplæg indenfor temaet: “Microdatamatbaseret måling/analyse/styring af et fysisk system”. - Projektet omfatter følgende fagområder: Digitalteknik, Analogteknik, Microcontrollerteknik, Mekanisk fysik. Metodisk: - Anvendelse af projektstyring i henhold til en faseorienteret projektmodel. - Anvendelse af studiekredse og selvstudie i forbindelse med tilegnelse af ny viden.

Synopsis (in Danish): Fagligt: - Gennemførelse af et projekt med udgangspunkt i et projektoplæg indenfor temaet: “Specifikation og syntese af hardware til formålsbundne systemer”. - Projektet omfatter følgende fagområder: Digitalteknik, Hardwarebeskrivende sprog, Hardwaresyntese, Programmerbar logik, Mikroprocessorsystemer. Metodisk: - Anvendelse af projektstyring i henhold til en faseorienteret projektmodel. - Anvendelse af studiekredse og selvstudie i forbindelse med tilegnelse af ny viden.

Synopsis (in Danish): Fagligt: - Gennemførelse af et projekt med udgangspunkt i et projektoplæg indenfor temaet: “Udvikling af systemsoftware til indlejrede systemer”. - Projektet omfatter følgende fagområder: Operativsystemer: Procesadministration, -skedulering, -synkronisering, proceskommunikation. Systemprogrammering. Datakommunikation og kommunikationsprotokoller. Metodisk: - Anvendelse af projektstyring i henhold til en faseorienteret projektmodel. - Anvendelse af studiekredse og selvstudie i forbindelse med tilegnelse af ny viden.

Synopsis (in Danish): Fagligt: - Gennemførelse af et projekt med udgangspunkt i et projektoplæg indenfor temaet: "Styring, regulering og overvågning". - Projektet omfatter følgende fagområder: Digital signalbehandling, Datakonvertering, Sensorer, aktuatorer og interfacing, Regulering.Metodisk: - Anvendelse af projektstyring i henhold til en faseorienteret projektmodel. - Anvendelse af studiekredse og selvstudieer i forbindelse med tilegnelse af ny viden. - Dokumentation.

Synopsis (in Danish): Low-level programmeringsteknikker. Programmering i assembler og C. Udviklingsmiljø: Editor, assembler, compiler, target-system, debygger.
Periferikredse (programmering med udgangspunkt i databladet.). Operativsystem struktur. Proces administration: Processer, CPU scheduling, Proces synkronisering, Interproces kommunikation, Deadlocks. Lager administration: RAM-lager administration, Virtuel hukommelse. Filsystem. I/O-system. Sikkerhed. Introduktion til distribuerede systemer.

Synopsis (in Danish): Grundlæggende ledningsteori: Model af transmissionsledning. Laplacetransformation. Transmissionsligninger, refleksion, ledningsimpedans. Puls på ledning. Refleksionsdiagram. Datakommunikation: Kodning og modulationsformer. Synkroniseringsmetoder. Fejldetektion og -korrektion. Kommunikationsnetværk: Referencemodeller. Netværkstopologier. Adresserings- og rutningsprincipper. Rutnings- og transportprotokoller.

Synopsis (in Danish): Centrale begreber fra tidskontinuert systemteori: Overføringsfunktion, Frekvensanalyse. Datakonvertering: Ideel og praktisk sampling og rekonstruktion, Aliasing, Kvantisering og dynamikområde, Konverteringsprincipper (A/D, D/A), Implementationsprincipper (Sample & Hold, A/D, D/A), Multirate sampling. Digital signalbehandling: Diskret-tid signaler og systemer i tids- og frekvens-domænet, Z-transformationen, Overføringsfunktion for lineære tidsinvariante systemer, Systemanalyse, Frekvensanalyse, Lineær fase systemer, Realisationsstrukturer for diskret-tid systemer. Hovedanvendelser af digital signalbehandling: Digitale IIR-filtre og transformation af analoge filtre, Digitale FIR-filtre og vindues-funktioner, Diskret Fourier-transformation og Fast Fourier-transformation, Adaptive filtre. Introduktion til beregnings/simulerings-værktøjer.

Synopsis (in Danish): Matematiske og grafiske metoder til syntese af lineære tidsinvariante systemer. - modellering. - linearisering. - diskret og kontinuert overføringsfunktion. - diskret og kontinuert tilstandsbeskrivelse. - analyse i tid og frekvens. - stabilitet, reguleringshastighed, følsomhed og fejl. - digitale PI, PID, LEAD og LAG regulatorer (serie regulatorer). - tilstandsregulering, pole-placement og tilstands-estimering (observer). - optimal regulering (least squares) og optimal tilstands-estimation (kalman filter). - realisering af digitale regulatorer. - introduktion til adaptiv og fuzzy regulering. - anvendelse af numeriske simuleringsværktøjer.