|
|
System for Automatic Calibration of a Robotic Tool
Šála, David ; Chromý, Adam (referee) ; Žalud, Luděk (advisor)
This Master's thesis describes the design and implementation of an experimental sample for automatic calibration of a robotic tool using machine vision methods under the auspices of the company SANEZOO EUROPE s.r.o. It deals with the analysis of all used methods of performing TCP calibration, on the basis of which it is implemented. The application is based on the Point-counterpoint method, where the robot is guided against the calibration point from three different directions, all perpendicular to each other. The calibration point is set using the ArUco marker. In order to detect the endpoint are used images from two cameras that are at the right angles to each other. Using conventional computer vision methods and an HSV filter, the endpoint of the instrument is found in the images and is guided to the calibration point. From the obtained coordinates, the searched endpoint of the robotic tool in the robot coordinates is found using the optimization method Particle Swarm Optimization. This application, therefore, performs TCP calibration in a fast time, thus reducing production downtime almost without human intervention.
|
|
|
Robot path planning by means of swarm intelligence
Schimitzek, Aleš ; Krček, Petr (referee) ; Dvořák, Jiří (advisor)
This diploma thesis deals with the path planning by swarm intelligence. In the theoretical part it describes the best known methods of swarm intelligence (Ant Colony Optimization, Bee Swarm Optimization, Firefly Swarm Optimization and Particle Swarm Optimization) and their application for path planning. In the practical part particle swarm optimization is selected for the design and implementation of path planning in the C#.
|
|
|
Evolutionary Algorithms for Neural Networks Learning
Vosol, David ; Rozman, Jaroslav (referee) ; Zbořil, František (advisor)
Main point of this thesis is to find and compare posibilities of cooperation between evolutionary algorithms and neural network learning and their comparison with classical learning technique called backpropagation. This comparison is demonstrated with deep feed-forward neural network which is used for classification tasks. The process of optimalization is via search of optimal values of weights and biases within neural network with fixed topology. We chose three evolutionary approaches. Genetic algorithm, differential evolution and particle swarm optimization algorithm. These three approaches are also compared between each other. The demonstrating program is implemented in Python3 programming language without usage of any third parties libraries focused on deep learning.
|
|
|
Determination of input parameters of advanced soil constitutive models using optimization methods
Koudela, Pavel ; Miča, Lumír (referee) ; Chalmovský, Juraj (advisor)
Choice of the constitutive model and determination of input parameters are necessary for a successful application of numerical methods in geotechnical engineering. Higher complexity of modern constitutive models results in an increase of the number of input parameters and time requirements for their calibration. Optimization methods are a possible solution for this problem. An application in which metaheuristic optimization method Particle swarm optimization (PSO) is involved is presented in this thesis. Critical review and testing of various PSO alternatives was performed in the first part of this thesis. The most efective PSO alternatives were chosen. In the second part connection between PSO algorithm and finite element solver was prepared. Automatization of determination of constitutive models input parameters was performed on three boundary value problems: laboratory test (oedometer), in-situ test (pressuremeter) and geotechical construction (retaining wall). Three types of constitutive models are used. Linear elastic-perfectly plastic Mohr-Coulomb model, elastoplastic non-linear Hardening soil model and Hardening soil - small strain model.
|
|
|
Evolutionary Design of Quantum Operator
Kraus, Pavel ; Mrázek, Vojtěch (referee) ; Bidlo, Michal (advisor)
The goal of this thesis is to utilize various evolutionary algorithms for quantum operator design in the form of unitary matrices in direct representation. Evolution strategy, differential evolution, Particle Swarm Optimization and artificial bee colony algorithms were chosen. In this thesis, the third and fourth algorithms were used for the first time in relation to quantum operator design. The experiments have shown that the utilization of direct representation gives results of acceptable quality.
|
|
|
Optimization of a Racing Car Setup within TORCS Simulator
Srnec, Pavel ; Jaroš, Jiří (referee) ; Pospíchal, Petr (advisor)
This master's thesis is about nature optimalization technigues. Evolution algortihms together with main thesis topic, Particle Swarm Optimization, is introduced in the following chapter. Car setup and simulator TORCS are introduced in next chapter. Design and implementation are introduced in next chapters. Destination of t master's thesis is finding optimal car setups for different curcuits.
|
|
|
Modelling and Assessment of Driver Behaviour Using Fuzzy Logic
Radvanský, Martin ; Mihálik, Ondrej (referee) ; Jirgl, Miroslav (advisor)
The work contains a research of mathematical modeling with a focus on the driver's behavior while driving. The following text contains the basic statistical processing of data obtained from the driving simulator for a total of 38 drivers. Subsequently, possible approaches to the use of fuzzy systems for modeling driver behavior are proposed. The main part of the work is focused on the verification of three hypotheses related to the use of fuzzy inference systems to model driver behavior.
|
|
|
Modeling of Linear Distortion of Audio Devices
Vrbík, Matouš ; Sysel, Petr (referee) ; Schimmel, Jiří (advisor)
Methods used for correction and modeling of frequency response of sound devices are discussed in this paper. Besides classic methods of digital filter design, more advanced and complex numerial methods are reviewed, Prony and Steiglitz-McBride in particular. This paper focuses on structure utilizing parallel sections of second-order IIR filters. Methods for calculating coefficients of this structure are presented and later implemented. For selected method, utilizing dual frequency warping, an interative algorithm for automatic calculation of parameters necessary to filter design is implemented - so called Particle Swarm Optimization. Six ways of evaluation filter design precision are presented and the results are compared. Functions realizing filter design are implemented in C++, MATLAB and Python. A VST module simulating the filter in real time is also provided.
|
|
|
Shape optimization of the typical frame of the load-bearing structure of the steel hall
Kuzbová, Šárka ; Vlk, Zbyněk (referee) ; Hokeš, Filip (advisor)
The thesis deals with the shape optimization of a typical frame of a steel hall structure. The calculation is performed for two types of parametric models created with both 1D and 2D finite element models in RFEM 6. The aim of the optimization process is to achieve a state where the structure has the lowest mass while satisfying the stress limit. The first model, which is created using 1D finite elements, focuses on optimizing the initial and final section heights of the members with the raises. The following model, which is developed using 2D finite elements, is aimed at optimizing the openings of the pierced beam. The thesis also includes analytical verification of the model using the deformation method.
|
|
|
Prediction of Multiple Time Series at Stock Market Trading
Palček, Peter ; Zbořil, František (referee) ; Rozman, Jaroslav (advisor)
The diploma thesis comprises of a general approach used to predict the time series, their categorization, basic characteristics and basic statistical methods for their prediction. Neural networks are also mentioned and their categorization with regards to the suitability for prediction of time series. A program for the prediction of the progress of multiple time series in stock market is designed and implemented, and it's based on a model of flexible neuron tree, whose structure is optimized using immune programming and parameters using a modified version of simulated annealing or particle swarm optimization. Firstly, the program is tested on its ability to predict simple time series and then on its ability to predict multiple time series.
|
guest ::
