March 15-17, 2017 – Industrial Research Institute for Automation and Measurements (PIAP).
3 keynote talks and 61 regular papers in 4 topic sessions.
Titles and abstracts of the papers ordered by topic session are presented below.
1.
Petri Nets in Discrete-Event and Hybrid Systems Modelling, Analysing, Performance Evaluation and Control
František Čapkovič
Institute of Informatics, Slovak Academy of Sciences
Frantisek.Capkovic@savba.sk
Abstract: Place/transition Petri nets, timed Petri nets (here time expresses the duration of separate operations) are used here in order to model, analyse, evaluate performance, and control discrete-event systems. Typical representatives of discrete-event systems are flexible manufacturing systems, transport systems, communication systems, etc. Hybrid Petri nets, more precisely first-order hybrid Petri nets, are used here in order to model, analyse and control hybrid systems. Typical representatives of hybrid systems are manufacturing systems in general – industrial production systems. The systems in question usually consist of cooperating subsystems (modules, agents). Particular approaches are illustrated on four case studies.
2.
Mobile robot localization: where we are and what are the challenges?
Piotr Skrzypczyński
Poznań University of Technology
piotr.skrzypczynski@put.poznan.pl
Abstract: This article surveys recent developments in the area of mobile robot localization. The focus is on indoor 3-D localization from vision and RGB-D data. We analyze three important aspects of the architecture of localization systems: perception, representation of the obtained data, and estimation of the robot trajectory from the internal representation of the outer environment. We attempt also to identify challenges and open problems in the domain. The analysis is illustrated by extensive references to the selected literature, as this paper was also conceived as a guide for those researchers, who want to enter the fascinating realm of SLAM for the first time.
3.
Napęd elektryczny w sterowaniu ruchem – rozwiązania i wyzwania
Teresa Orłowska-Kowalska
Wrocław University of Science and Technology
teresa.orlowska-kowalska@pwr.edu.pl
Abstract: Abstract is available in Polish only.
4.
The PLC implementation of Fractional -Order operator using CFE approximation
1Krzysztof Oprzędkiewicz, 1Wojciech Mitkowski, 2Edyta Gawin
1AGH University of Science and Technology, 2State Higher Vocational School in Tarnów
kop@agh.edu.pl, wojciech.mitkowski@agh.edu.pl, e_gawin@pwsztar.edu.pl
Abstract: In the paper an implementation of an elementary fractional order, integro-differential operator at PLC platform is discussed. The considered element is approximated with the use of discrete CFE approximation. The operator we deal with is a crucial part of fractional order PID controller. Guidelines to PLC implementation with the use of object-oriented approach presented by IEC 61131.3 standard are given also. As an example the implementation at SIEMENS SIMATIC S7 1200 platform is presented. As a reference the analytical response of element was applied, the quality of model was estimated with use of typical MSE cost function. Results of experiments show, that the PLC implementation of the fractional order element is possible to make with the use of object-oriented approach and the accuracy of approximation is determined by its order.
5.
An accuracy estimation for a non integer order, discrete, state space model of heat transfer process
1Krzysztof Oprzędkiewicz, 1Wojciech Mitkowski, 2Edyta Gawin
1AGH University of Science and Technology, 2State Higher Vocational School in Tarnow
kop@agh.edu.pl, wojciech.mitkowski@agh.edu.pl, e_gawin@pwsztar.edu.pl
Abstract: In the paper an accuracy analysis for non integer order, discrete, state space model of heat transfer process in one dimensional plant is presented. The proposed model is a discrete version of time – continuous, non integer order, state space model proposed previously by Authors. The discretization of integro/differential operator was done with the use of backward difference method. The accuracy and convergence of the discussed model was considered as a function of model order and memory length necessary to proper estimation of non integer order order operator. Results of experiments show that the proposed, discrete model assures the good performance in the sense of MSE cost function, but its size is relatively high.
6.
Initial Comparison of Experimental vs.~Simulation Results of Fractional-Order PI Controller
1Talar Sadalla, 2Dariusz Horla
1Institute of Control and Information Engineering, Poznań University of Technology, 2Poznań University of Technology
talar.h.sadalla@doctorate.put.poznan.pl, dariusz.horla@put.poznan.pl
Abstract: In the paper, an initial comparison of experimental vs.~simulation results from a tracking system is presented, in the case of a Fractional-Order PI controller for a time-delay system. The controller is implemented as to work in real-time regime, for the Modular Servo System (Inteco). Based on theoretical results, stability regions are computed using Hermite-Biehler and Pontryagin theorems. Next, based on identification carried out in previous work, simulation results of a tracking performance are presented, and compared with experimental results from the laboratory stand, to verify if any conclusions can be drawn using computer models, concerning real-world control system with fractional-order controller. To compare control performance, IAE and ISE indices are used.
7.
Identification of parameters of the capacitor equivalent scheme using Monte Carlo methods
1Stefan Kubisa, 2Zygmnt Warsza
1emeritus prof. of West Pomeranian University of Technology Szczecin, 2Industrial Research Institute for Automation and Measurements (PIAP)
dskubisa@hotmail.com, zlw1936@gmail.com
Abstract: The identification of the internal parameters of the model of real capacitor using Monte Carlo (MC) method is considered. It is based on measurement results of the capacitor impedance or admittance components available for selected frequencies on its terminals. The measured parameters and identified parameters are usually linked by a system of nonlinear relationships and their analytical solution is either very troublesome or even non-existent. As an example, parameters of the five-element equivalent circuit of the real capacitor are identified by two Monte Carlo (MC) methods, single and multi-iterative. The results of simulated measurements of the equivalent impedance components of capacitor at several frequencies are given. From the received distributions of searched parameters, the accuracy of their identification is estimated. Obtained results show that the identification performed by MC multi-iterative method is sufficiently fast and is possible to obtain an acceptable accuracy. Then this MC method is recommended for use in the measurement practice.
8.
Descriptor positive nonlinear systems
Tadeusz Kaczorek
Bialystok University of Technology
kaczorek@isep.pw.edu.pl
Abstract: The positivity of discrete-time and continuous-time nonlinear systems is addressed. Necessary and sufficient conditions for the positivity of the descriptor nonlinear systems are established. A procedure for checking the positivity is proposed and demonstrated on numerical examples.
9.
Descriptor positive nonlinear systems
Lukasz Sajewski
Bialystok University of Technology
l.sajewski@pb.edu.pl
Abstract: Descriptor fractional continuous-time linear systems are addressed. Three different methods for finding the solution to the state equation of the descriptor fractional linear system are considered. The methods are based on: Shuffle algorithm, Drazin inverse of the matrices and Weierstrass-Kronecker decomposition theorem. Effectiveness of the methods is demonstrated on simple numerical example.
10.
Full order observer synthesis for singular discrete-time fractional systems
Rafał Kociszewski
Bialystok University of Technology
r.kociszewski@pb.edu.pl
Abstract: The paper is devoted to observer synthesis for linear singular discrete-time fractional systems. The problem of finding a nonnegative gain matrix of the observer such that the observer is asymptotically stable is formulated and solved by the use of linear matrix inequality (LMI) method. The proposed approach to the observer synthesis is illustrated by theoretical example.
11.
Pointwise completeness and pointwise degeneracy of selected class of singular linear discrete-time systems
Rafał Kociszewski
Bialystok University of Technology
r.kociszewski@pb.edu.pl
Abstract: The paper presents a problem of pointwise completeness and pointwise degeneracy of selected class of singular linear discrete-time systems. It has been shown that after decomposition of considered system into two independent systems: regular (standard) fractional system and closely singular system (with a nilpotent matrix N) pointwise completeness and pointwise degeneracy conditions can be formulated in reference to standard fractional discrete-time system. Proposed approach is possible if the matrix N = 0. The considerations are illustrated by a numerical example.
12.
Robust stability of a class of an uncertain fractional discrete-time linear state-space system
Andrzej Ruszewski
Faculty of Electrical Engineering, Bialystok University of Technology
a.ruszewski@pb.edu.pl
Abstract: This paper presents the problems of robust asymptotic stability of fractional-order discrete-time linear systems with uncertainty. It is assumed that the system matrix is the interval matrix and the fractional order alpha satisfies 0 < alpha < 1. The new robust stability conditions are given based on the matrix measure and Gershgorin’s theorem for the interval matrices. The considerations are illustrated by numerical examples.
13.
Comparison of two different methods of observer synthesis for descriptor discrete-time linear systems
Kamil Borawski
Bialystok University of Technology
kam.borawski@gmail.com
Abstract: Descriptor observers for descriptor discrete-time linear systems are investigated. Two different methods of observer synthesis are analyzed and the comparison of the methods is performed. The methods are based on: [Ez-A]^-1 expansion and Weierstrass-Kronecker decomposition. Considerations are illustrated by numerical examples.
14.
Realisation of Continuous-Time (Fractional) Descriptor Linear Systems. New Digraph-Based Method.
Konrad Markowski
Faculty of Electrical Engineering, ISEP, Warsaw University of Technology
Konrad.Markowski@ee.pw.edu.pl
Abstract: In this paper, a new digraph-based method for computation of a minimal realisation of a given proper transfer function of continuous-time (fractional) descriptor one-dimensional linear systems has been presented. For the proposed method, an algorithm was constructed. The proposed solution allows minimal digraphs construction for any one-dimensional system. The proposed method was discussed and illustrated with numerical examples.
15.
Relations Between Digraphs Structure and Analogue Realisations with an Example of Electrical Circuit
Konrad Markowski
Faculty of Electrical Engineering, ISEP, Warsaw University of Technology
Konrad.Markowski@ee.pw.edu.pl
Abstract: This paper present a method for computation of a minimal realization of a given proper transfer function of continuous-time fractional linear systems in the electrical circuit. For the proposed method, a digraph-based algorithm was constructed. We have also shown how after using the constant phase element method we can realize such a system. The proposed method was discussed and illustrated with some examples.
16.
Synthesis of optimal robust regulator for food processing facilities
1Natalya Lutskaya, 2Igor Korobiichuk, 1Anatoliy Ladanyuk, 1Serhii Naku, 2Maciej Kachniarz, 3Michał Nowicki, 3Roman Szewczyk
1National University of Food Technologies, Ukraine, 2Industrial Research Institute for Automation and Measurements (PIAP), 3Institute of Metrology and Biomedical Engineering, Warsaw University of Technology
lutskaya@yandex.ru, kiv_Igor@list.ru, ladanyuk@ukr.net, naku.serega@gmail.com, mkachniarz@piap.pl, nowicki@mchtr.pw.edu.pl, rszewczyk@piap.pl
Abstract: Food processing facilities operate under permanent perturbations. Their nature may be different: change of facility input load; external facility perturbations that change within wide limits; interference in communications lines; in-service evolution of internal facility parameters. These factors lead to a shift of optimum regulator settings and, consequently, deterioration in the quality of transients in regulation system. If perturbation uncertainty range is substantial, control system should imply modification of regulator setting, i.e. development of adaptive system. However, there is another approach, in which regulator settings are not changed. It is called a robust control system. Today there are various methods developed for determining a robust regulator, but they have significant drawbacks limiting and slowing their use in industrial environments, e.g. high-order regulator. The synthesis of matrix linear regulator for multivariable systems based on minimizing H∞-test has all the advantages of robust regulator and reduces interaction of contours. The paper studies and simulates an optimal robust control system for evaporation plant of sugar factory with a linear array regulator based on LMI-approach.
17.
LD Graphic Editor Implemented in CPDev Engineering Environment
Dariusz Rzońca, Jan Sadolewski, Andrzej Stec, Zbigniew Świder, Bartosz Trybus, Leszek Trybus
Rzeszów University of Technology
drzonca@prz.edu.pl, jsad@prz.edu.pl, astec@prz.edu.pl, swiderzb@prz-rzeszow.pl, btrybus@prz.edu.pl, ltrybus@prz.edu.pl
Abstract: Process of creating a diagram in LD graphical language in CPDev environment, translation to ST textual program, compilation into VMASM executable code and, finally, simulation is overviewed by means of typical example. Earlier version of CPDev LD editor did not suit industrial applications, particularly with respect to processing other variables than boolean. Basic functionalities of LD in CPDev are now not far apart from those of renown CODESYS package.
18.
SFC Graphic Editor for CPDev environment
Andrzej Stec
Rzeszów University of Technology
astec@prz.edu.pl
Abstract: Process of creating a diagram in a new SFC graphical editor of the CPDev environment, as well as translation to ST textual language, compilation into VMASM executable code, and simulation in CPSim tool are presented. Basic functionalities of the SFC editor are now similar to those of renowned commercial software.
19.
Flight trajectory from the selected class of automatically controlled evasive manoeuvres in case of a moving obstacle
Jerzy Graffstein
Institute of Aviation
jerzy.graffstein@ilot.edu.pl
Abstract: Owing to flight safety, the shape of trajectory of the evasive manoeuvre, performed to avoid a moving obstacle, is of important meaning. In the article the method is proposed for determining the shape of flight trajectory contained in a selected class of complex evasive manoeuvres. In the course of the process the following phases of motion are identified: the collision avoidance phase, the passing by the obstacle phase, the phase of returning to the trajectory pre determined in flight plan. The complex shaped segments of flight trajectory are assigned to these phases. The method capable to identify conditions of the occurrence of collision threat is presented. To define it, selected state variables of the aircraft and obstacle system and relationships linking them up, are used. Numerical flight simulations are completed to cover the aforementioned phases of evasive manoeuvre and selected scenarios of objects motion. Selected results of numerical investigations are presented and discussed
20.
Automatic parallel parking of a car – theoretical considerations and simulation studies
Maciej Trojnacki
Industrial Research Institute for Automation and Measurements (PIAP)
mtrojnacki@piap.pl
Abstract: The paper is concerned with the problem of automatic parallel parking of a car. Current state of the art in self-driving cars and methods of automatic parking were discussed. For simulation studies the kinematic bicycle model of the vehicle was assumed. A method of solving the problem of automatic parallel parking was proposed. The idea of this method consists in planning of characteristic poses of the vehicle and realization of motion between those poses using pose controller. Results of simulation research were presented and discussed, which confirmed the correctness of the approach. After appropriate modification of the proposed solution in terms of planning of poses, this method can be also used to perform other parking maneuvers.
21.
Application of bio signals in the brain-device interfaces
Arkadiusz Kubacki, Andrzej Milecki
Poznań University of Technology
arkadiusz.j.kubacki@doctorate.put.poznan.pl, andrzej.milecki@put.poznan.pl
Abstract: In the paper, a review of the current state of knowledge in the field of brain-machine interfaces based on bio signals, is presented. Different types of bio sig-nals, which can be used to control of devices are shortly characterized. Special at-tention is given to the use of EEG signals in the most interesting and complex brain-machine interfaces. The EEG device is described and characterized. Initial investigation results made at Poznan University of Technology in the area of ap-plication of consumer type EEG devices to brain signal acquisition are presented. The use of artificial neural network for this signal recognition are described and discussed.
22.
Development of electronic controller for haptic joystick and electrohydraulic drive
Dominik Rybarczyk, Piotr Owczarek, Arkadiusz Jakubowski
Poznań University of Technology
dominik.rybarczyk@put.poznan.pl, piotr.owczarek@put.poznan.pl, arkadiusz.z.jakubowski@doctorate.put.poznan.pl
Abstract: The article describes design of control electronic for one axis haptic joystick. System based on 32-bits microcontroller equipped with DMA controller and 12-bits ADC inputs. Authors implemented on-off current regulator based on the information taken from the hall effect sensor. The built electronic was tested with use of an electrohydraulic servodrives controlled by a proportional valve. Authors collected step responses of a current and a force during several experiments. Performed tests is a start point for built up electronic controller for hydraulic manipulator in a hydraulic lift car.
23.
Dynamic model and simulation of electrohydraulic proportional valve
Piotr Owczarek, Dominik Rybarczyk, Arkadiusz Kubacki
Poznań University of Technology
piotr.owczarek@put.poznan.pl, dominik.rybarczyk@put.poznan.pl, arkadiusz.j.kubacki@doctorate.put.poznan.pl
Abstract: In the paper, a modelling of electrohydraulic servo drive is presented. The authors proposed implementation of a dynamic model of proportional valve with the one of most important non linearity, which is square root flow characteristic. Following model is useful when system directly has got information about position of slider in valve. Authors proposed to extend that model to electro-mechanic parts with dedicated control cards. An input voltage of real valve in range ±10V was converted to position of slider by use of transfer function with transport delay and dead zone. The result of simulation and comparison with real object were presented. Similarities model with valve for complete model was 94% (in the worst case: for step response with input signal 2V). The best results was achieved for step response of 10V, with similarities (accuracy) 99.42%.
24.
Designing mass-customized network of passenger services subject to grid topology constraints
Grzegorz Bocewicz, Zbigniew Banszak
Koszalin University of Technology
bocewicz@ie.tu.koszalin.pl, zbigniew.banaszak@tu.koszalin.pl
Abstract: Using fractal organization perspective, this paper provides an unified approach to multimodal routing and timetables synchronization problems imposed by public transport services within city’s grid-like network. The goal is to provide a declarative modelling driven approach enabling assessment of the right match-up of local transportation line schedules to a given passenger flow itineraries. Assuming city’s grid-like clusters resulting in mesh-like street networks the corresponding transportation network composed of a set of same pattern repetitively acting local transportation modes is considered. In that model the passenger travel schedules can be estimated easily while taking into account cyclic behaviour of both: local transportation modes and the whole transportation network. A numerical case is presented to demonstrate the feasibility of the proposed approach, and the sensitivity of the grid-like structure of transportation network with respect to their spatial features such as heterogeneity and symmetry is also examined
25.
A hybrid CLP/MP approach to modeling and solving resource-constrained scheduling problems with logic constraints
Paweł Sitek, Jarosław Wikarek, Tadeusz Stefanski
Kielce University of Technology
sitek@tu.kielce.pl, j.wikarek@tu.kielce.pl, t.stefanski@tu.kielce.pl
Abstract: Constrained scheduling problems are common in everyday life and especially in: distribution, manufacturing, project management, logistics, supply chain management, software engineering, computer networks etc. A large number of integer and binary decision variables representing the allocation of dif-ferent constrained resources to activities/jobs and constraints on these decision variables are typical elements of the resource-constrained scheduling problems (RCSPs) modeling. Therefore, the models of RCSPs are more demanding, particularly when methods of operations research (OR) are used. By contrast, most resource-constrained scheduling problems can be easily modeled as instances of the constraint satisfaction problems (CSPs) and solved using constraint logic programming (CLP) or others methods. Moreover, CLP-based environments enable easy modeling of various types of constraints including logic constraints. In the CLP-based environment the problem definition is separated from the algorithms and methods used to solve the problem. Therefore, a hybrid approach to resource-constrained scheduling problems that combines an OR-based approach for problem solving and a CLP-based approach for problem modeling is proposed. To evaluate the efficiency and flexibility of this ap-proach, illustrative examples of resource-constrained scheduling problems with logic constraints are implemented using hybrid CLP/MP approach.
26.
Methods of determining information support of web community user personal data verification system
1Solomia Fedushko, 1Yuriy Syerov, 2Igor Korobiichu, 3Andrzej Juś
1Lviv Polytechnic National University, 2Institute of Automatic Control and Robotics, Warsaw University of Technology, 3Industrial Research Institute for Automation and Measurements (PIAP)
felomia@gmail.com, syerov@ridne.net, kiv_igor@list.ru, ajus@piap.pl
Abstract: This article considers the concept of information support of web community member personal data verification system. It has been established that information support of verification system depends on the following parameters: profile filling level, profile data and content actuality level, administrative authority, web user activity level and level of compliance with virtual community rules. The paper also analyzes an actual task of developing methods for determining information support of web community member personal data verification system. The level of information support of web community member personal data verification system allows evaluating the effectiveness of verification system in web-community management. It is concluded that the higher is the level of information support of web community member data verification system, the better is the community management performance, and therefore, web community operation.
27.
Experimental research of electrochemical energy storage
Adrian Chmielewski, Jędrzej Mączak, Przemysław Szulim
Institute of Vehicles, Faculty of Automotive and Construction Machinery Engineering, Warsaw University of Technology
a.chmielewski@mechatronika.net.pl, jma@mechatronika.net.pl, p.szulim@mechatronika.net.pl
Abstract: The paper presents experimental research involving VRLA (Valve Regulated Lead Acid) AGM (Absorbed Glass Mat)batteries. Test-bench research was conducted in the conditions of constant load current. The paper presents the temperature increase on the battery’s terminals and body accompanying battery discharge in the condi-tions of a preset ambient temperature. The paper also presents the influence that various discharge current values had on growth of the temperature rec-orded on the battery’s terminals as well as the change of voltage on the bat-tery’s terminals. Furthermore, the paper includes examination of the influ-ence that changes of ambient temperature have on change of a battery’s use-ful capacity. The influence of the changes of ambient temperature was ex-amined in a climatic chamber. Change of the battery’s internal resistance and the electromotive force, depending on the level of battery charging, are also presented. The analyses were conducted for a typical operating range of a electrochemical battery.
28.
Experimental research and simulation model of electrochemical energy stores
Adrian Chmielewski, Jędrzej Mączak, Przemysław Szulim
Warsaw University of Technology
a.chmielewski@mechatronika.net.pl, jma@mechatronika.net.pl, p.szulim@mechatronika.net.pl
Abstract: The first part of the paper presents the results of experimental research involving VRLA (Valve Regulated Lead Acid) AGM (Absorbed Glass Mat) batteries, the lithium-ion batteries and lithium iron phosphate (LiFePO4) batteries. The experimental research was conducted in a static cycle (with constant load current). The paper presents the temperature increase on the battery’s terminals and body. The influence that various values of discharge current have on growth of temperature and change of voltage on the battery’s terminals is also presented. The second part of the paper contains the analytical relations which have been used for building the simulation model in the Matlab&Simulink environment. The results obtained on the basis of the model have been validated against the results of experimental research.
29.
Path Planning for an Unmanned Ground Vehicle Traversing Rough Terrain with Unknown Areas
Piotr Skrzypczynski
Poznań University of Technology
piotr.skrzypczynski@put.poznan.pl
Abstract: In this paper we tackle the problem of planning a path for a ground vehicle in rough terrain, which is perceived and modelled in an imperfect way. The imperfectness, manifesting itself as unknown areas decreases the efficiency of motion planning, as the unknown terrain has to be treated as non-traversable. We demonstrate that it is possible to employ relatively simple image inpainting algorithms to substitute the missing elevation values in the terrain map, and to embed the terrain dis- continuity filling procedure in an efficient path planner. We show reliable filling-in of discontinuities in various elevation maps, and then demon- strate gains in path planning efficiency due to integration between the discontinuity filling procedure and the path planner.
30.
Development of vertical movement controller for multirotor UAVs
Przemysław Gąsior, Adam Bondyra, Stanisław Gardecki
Poznań University of Technology
przemyslaw.gasior@cie.put.poznan.pl, adam.bondyra@put.poznan.pl, stanislaw.gardecki@put.poznan.pl
Abstract: In this paper, a vertical velocity controller for multirotor UAVs is proposed. Based on the previous research, authors developed a model of vertical movement which takes into consideration measurement noises, and designed a simulation that allowed tuning of mentioned controller. As a regulation scheme, a classical PI structure was used. The derivative part was neglected because of high amplitude of noise during harsh touchdowns. Tuning of parameters was achieved by PSO optimisation. Experimental results showed that the selected control structure and its parameters fulfill stated requirements. In addition, developed simulation is adequate to the real platform.
31.
Multirotor aerial platform with manipulation system – static disturbances
Stanisław Gardecki, Andrzej Kasiński, Adam Bondyra, Przemysław Gąsior
Poznań University of Technology
stanislaw.gardecki@put.poznan.pl, andrzej.kasinski@put.poznan.pl, adam.bondyra@put.poznan.pl, przemyslaw.gasior@cie.put.poznan.pl
Abstract: Adding on-board manipulation system to flying robot rises a serious problem , as its operation strongly affects the overall stability of the aerial platform. With arms of the manipulation system moving during flight, the distribution of masses changes and extra dynamic reaction forces are generated. Such disturbances are usually not taken into consideration in the standard platform stabilization algorithms. This is why a new algorithm addressing such disturbances is necessary. Manipulators perform various movements, e.g. symmetric or independent. Therefore, it is necessary to evaluate the induced disturbances, both in qualitative and quantitative aspect. In this paper, the static reactions due to the changes of the system’s configuration during flight are investigated.
32.
Experimental test bench for multirotor UAVs
Adam Bondyra, Przemysław Gąsior, Stanisław Gardecki
Poznań University of Technology
adam.bondyra@put.poznan.pl, przemyslaw.gasior@cie.put.poznan.pl, stanislaw.gardecki@put.poznan.pl
Abstract: Development of methods of state estimation and control for multirotor micro UAVs is often a long and troublesome process. In many cases, various devices that allow to test and evaluate solutions in safe, controllable environment are used. This paper presents a design of custom test bench for unmanned multirotor aerial vehicles. Developed device provides a intermediate step between simulation phase and test flights, where some of basic properties of flight can be evaluated. Thanks to fixing the vehicle on a triaxial gimbal, it is possible to simulate its behaviour during flight scenarios. Set of precise encoders, mounted on every axis, delivers ground truth data about vehicle’s attitude. Solutions that allow to precisely synchronize data from both sources, the test bench and multirotor’s on-board sensors, are presented. Custom PC software provides functionality of storing and presenting gathered data. The whole systems is a low-cost, easy to replicate solution, with simple components manufactured using FDM 3D printing technology. The test bench has been evaluated as helpful tool in development of various aerial platforms. In addition, the system can serve in teaching the basics of automatic control in aerial vehicles.
33.
CIE-DataGlove, a Multi-IMU System for Hand Posture Tracking
Tomasz Mańkowski, Jakub Tomczyński, Piotr Kaczmarek
Poznań University of Technology
tomasz.mankowski@put.poznan.pl, jakub.tomczynski@put.poznan.pl, piotr.kaczmarek@put.poznan.pl
Abstract: In this paper, a fully functional dataglove device, called CIE-DataGlove, is presented. CIE-DataGlove is a glove-like apparatus intended for human hand posture capture. The essential design requirement was not to hinder hand movement and object manipulation, and not to introduce additional mechanical resistance to the fingers. The system is based on 12 9-DOF inertial measurement units placed on phalanges and metacarpus. The article describes mechanical, hardware, and software implementations used in the development of CIE-DataGlove. Throughput performance of communication interface is studied in the context of interface response time.
34.
Optimized and Reconfigurable Evironment for Simulation of Legged Robots
1Dominik Belter, 1Mateusz Spis, 1Adam Matecki, 1Patryk Maik, 1Adam Kurzawa, 2Marek Kopicki
1Poznań University of Technology, 2University of Birmingham, UK
dominik.belter@put.poznan.pl, mspis@wp.pl, adam.p.matecki@gmail.com, patryk1.maik@gmail.com, adamkurzawa0@gmail.com, M.S.Kopicki@cs.bham.ac.uk
Abstract: The paper presents a reconfigurable simulator of legged robots. The simulator is based on the physics engines which model the motion of rigid bodies. The goal of this research is to design reliable tool for verification new control concepts for various types of legged robots. To this end, the new architecture of robot’s configuration scheme is proposed. The new hierarchical structure of the description files allows to re-use mechanical parts of existing robots and rapidly prototype new mechanical systems. We also propose the optimization method which increases the stability of the simulator. The simulator tuning technique allows to find the set of parameters and reduce the discrepancy between the simulated and the real robot.
35.
Influence of exoskeleton parameters calibration inaccuracies on the step length error
Rafał Kabaciński, Mateusz Kowalski, Piotr Kaczmarek
Institute of Control and Information Engineering, Poznań University of Technology
rafal.kabacinski@put.poznan.pl, mateusz.kowalski@doctorate.put.poznan.pl, piotr.kaczmarek@put.poznan.pl
Abstract: While exoskeletons for paraplegic users are used nowadays only in laboratory environments for therapy, their use in less orderly environments or for climbing stairs is still troublesome for operators. One of the factors is the high step length variance. In this paper, the analysis of the influence of exoskeleton parameters calibration inaccuracies on the sagittal step length error in regard to the different dimensions and programmed step lengths is performed on CIE~Exoskeleton kinematic model. Results show that the exoskeleton parameters calibration inaccuracies have a significant influence on the step length error. The exoskeleton segments lengths, have to be measured with at least millimeter accuracy, and special care is advised in case of the rigidity of the back segment of the exoskeleton.
36.
The estimation method of strenght for technology-oriented 3D printing parts of mobile robots
Maciej Cader
Industrial Research Institute for Automation and Measurements (PIAP)
mcader@piap.pl
Abstract: Application of Rapid Tooling technology for manufacturing of robotic parts be-came possible due to development of Fused Deposition Modeling method. The intensive progress in this method improvement was an effect of interdisciplinary cooperation of material science, physics and production engineers. FDM opened new horizons for many fields of e.g. industrial and medical applications. Produc-tion of customized demanding robotic parts requires effective method and reliable materials. The article presents results of estimation of strenght for the robotics part produced using FDM technology. Selection of prototypes of the research components – the grab’s arm and jaw – arose from a potential use of FDM technology in the pro-duction of these types of components. The arm and the grab are situated at the end of the kinematic chain of the robot’s manipulator. Thus, they should be light so that any additional load is put on the manipulator’s motors. This has a direct impact on the manipulator’s maximum working load. These types of components, apart from the least possible weight, should have relatively high resistance due to the fact that the robot can grab and lift objects up to 15 kg. The third specification of such objects is their non-standard shape. The arm and the grab must be suita-ble for carrying out different objects and therefore their non-standard dimensions and shapes make their mass or even ordinary production in standard technologies (for example machining) impossible.
37.
Comparison of a traditional control and a force feedback control of the robot arm during teleoperation
1Marcin Chciuk, 2Andrzej Milecki, 1Paweł Bachman
1Faculty of Mechanical Engineering, University of Zielona Góra, 2Institute of Mechanical Technology, Poznań University of Technology
m.chciuk@iibnp.uz.zgora.pl, andrzej.milecki@put.poznan.pl, p.bachman@iibnp.uz.zgora.pl
Abstract: The article describes the manual control system of robot arm using the joystick with magnetorheological brakes, during which the operator observed the robot work with the use of cameras. The construction of the investigations stand, system kinematics and control algorithm were described. In the final part of the article the results of scientific research for the two control variants: the classical one and with force feedback one were discussed and compared.
38.
Depth-map-based shape recognition of soft continuum manipulator body
Jan Fraś
Industrial Research Institute for Automation and Measurements (PIAP)
jfras@piap.pl
Abstract: Soft robotics is young and dynamically evolving field of scientific research. Soft robots are difficult to control and much more complex for modeling than traditional robots. They are, however, gaining more and more researchers’ interest due to their high potential. Soft robot body is made of flexible materials and it contains no rigid or discrete joints, its movement is generated by smooth and continuous deformation of the body. The softness of the body enables wide range of possible robot configurations which, on the other hand, are difficult to measure or predict. There were many shape detection approaches proposed so far, but most of them provide only discrete manipulator points position. Such a solution is not sufficient for some applications, since the manipulator body configuration in between the measured points is uncertain. In this paper Authors propose algorithm for continuous detection of the flexible manipulator shape. The algorithm is based on depth image provided by sensor such as Kinect. The depth image is processed in steps: detection of the manipulator central axis, the axis 3D shape reconstruction and orientation approximation for each point that lies on it. Numerical representation of the manipulator surface is also generated. The data obtained can be used for manipulator internal state correction including its input values and external disturbances.
39.
Data-driven video game agent pathfinding
Paweł Stawarz, Zbigniew Świder
Rzeszów University of Technology
p.stawarz@prz.edu.pl, swiderzb@prz.edu.pl
Abstract: This paper proposes a computationally inexpensive algorithm that uses player data to optimize NPC pathfinding in a competitive, multiplayer environment. Statistics gathered during matches are subjected to pattern analysis and used to modify edge values of the map graph. Utilizing data describing player habits enhances the AI’s odds against the player by letting it better adapt to the situation. Combining the data concerning individual players results in the possibility to react to multiple human opponents at once, maintaining the ability to adapt even after a large number of matches. In order to further improve control over the agent, two novel variables are introduced, increasing the ability to adapt to behavior that is unique in a particular match and giving more control over the risk the agent is willing to take. The findings of the hereby paper can be applied to any pathfinding algorithm that works with directed graphs.
40.
The development of PIAP Fenix mobile robot.
Tomasz Krakowka, Bartosz Stankiewicz
Industrial Research Institute for Automation and Measurements (PIAP)
tkrakowka@piap.pl, bstankiewicz@piap.pl
Abstract: The article presents selected technical issues associated with development of small UGV – PIAP Fenix. Topics related to the design of the drive system robot, robot weight and noise reduction were discussed.
41.
Coverage region for the bidimensional vector measurand
Paweł Fotowicz
Główny Urząd Miar (Central Office of Measures)
uncert@gum.gov.pl
Abstract: The approach for calculating the coverage region associated with the bidimensional measurand are presented. The calculation methods are based on an uncertainty propagation and a propagation of distributions. The measure of uncertainty associated with the scalar measurand is a coverage interval, but the measure of uncertainty associated with the vector measurand is a coverage region. The coverage interval is a special case of the coverage region, when the scalar output quantity is represented by a univariate measurement function. The coverage region represents the uncertainty of vector output quantity defined by multivariate measurement model.
42.
Calibration of scanning electron microscope with improved model of the silicon relief measure
1Anton Shantyr, 1Eugenij Volodarski, 2Zygmunt Lech Warsza
1National Ukraine University of Technology, Kijev Politechnic, 2Industrial Research Institute for Automation and Measurements (PIAP)
anton.shantyr@gmail.com, vet-1@ukr.net, zlw1936@gmail.com
Abstract: Accuracy of measurements with scanning electron microscope (SEM) depends on the environment conditions, noises in nanometer-ranges (caused by non-ideal surface of measures), non-ideal SEM design, influence of external factors (vibration, acoustic noise, magnetic fields, etc.). To reduce the influence of these variables, the calibration procedure, with the use of silicon measures to reproduce dimensions is performed preferably. The model of measure real image (which is used for calibration) can be formalized as the convolution of the level of secondary electron emission and the distribution of electrons in the crossover of the electron probe. Based on the developed model, the general approach to SEM calibration is formulated and proved. The step-by-step removal of non-parametric and parametric uncertainty of scale factor and electron probe equivalent diameter is proposed. As the first step, the approximation of individual cuts on the measured object image, which correspond to the same cuts of measure relief with further averaging and estimating the metrological parameters and their error distributions, is performed. In the second step (of optimal estimation), the vector containing metrological characteristics using Bayesian criterion and the theory of optimal estimations is find. Proposed approach can significantly increase SEM accuracy and performance
43.
A polynomial estimation of measurand parameters for samples of non-Gaussian symmetrically distributed data
1Zygmunt Lech Warsza, 2Serhii Zabolotnii
1Industrial Research Institute for Automation and Measurements (PIAP), 2Cherkasy State Technological University, Ukraine
zlw1936@gmail.com, zabolotni@ukr.net
Abstract: The non-standard method for evaluating of the average and standard deviation of the symmetrically non-Gaussian-distributed data of sample with a priori partial description (unknown PDF) is proposed. This method of statistical estimation is based on the apparatus of stochastic polynomials and uses the higher-order statistics (moment & cumulant description) of random variables. The analytical expressions for finding estimates for the degree of the polynomial S = 3 and their accuracy analyzes are given. It is shown that the uncertainty estimates received for polynomial are generally less than the uncertainty estimates obtained based on the mean (arithmetic average). Reduction factor, which depends on the MSE values of higher order cumulant coefficients, characterizes the degree of the sampling distribution differences from the Gaussian model. The results of statistical modeling, based on the Monte Carlo method, confirmed the effectiveness of the proposed approach are presented.
44.
Approach to determination of parameters of probability density function of object attributes recognition in space photographs is considered within statistical method
1Igor Korobiichuk, 2Ruslan Osadchuk, 2Dmytro Fedorchuk, 1Paweł Nowak
1Industrial Research Institute for Automation and Measurements (PIAP), 2Zhytomyr Military Institute n.a. S.P. Korolyov, Ukraine
kiv_Igor@list.ru, tpk2210@ukr.net, fedor4uk.d@gmail.com, pnowak@piap.pl
Abstract: High information richness of satellite images is used effectively, only if they are processed promptly. Analysis of recent research shows that existing hardware and software allows only partial automation of object recognition in space photographs. Automation is reduced to visualization of images and measurement of their parameters. Further processing requires parameters of probability density function of object features recognition. To solve certain problems in geological, hydrological, forestry and other types of decryption, parameters of this distribution are estimated according to experimental data. However, this approach is not suitable for recognition of single compact surface objects. Therefore, reference images formed upon three-dimensional models are suggested to determine unknown parameters of probability distribution function. Method of allowable transformations is applied to determine initial conditions of reference images and take their recognition feature distributive law for a distributive law of recognition features of single compact surface objects.
45.
Vision system for Inspection of Glass Furnace Structure
Piotr Garbacz, Tomasz Giesko, Piotr Czajka, Adam Mazurkiewicz
Institute for Sustainable Technologies – National Research Institute
piotr.garbacz@itee.radom.pl, tomasz.giesko@itee.radom.pl, piotr.czajka@itee.radom.pl, adam.mazurkiewicz@itee.radom.pl
Abstract: The article presents a method for the inspection of the walls and the bottom of a glass furnace in operation using a hybrid vision system that allows simultaneous monitoring in infrared and visible light. The fusion of images recorded in the infrared and visible light facilitates the process of the identification of specific areas on the surface of the furnace. Evaluation of the condition of the structure of the walls and bottom of the furnace is based on the analysis of thermograms recorded at specified intervals during the operation of the furnace. The paper presents the recorded thermal images and photographs of damage to the structure of the furnace. The collected results of the inspection of the furnace form the basis for the development of an empirical model to present the process of the degradation of the insulation layer of the furnace.
46.
Optimization Of Interpolation For Improved Numeric Calculation Of Forward Eddy Current Tomography Transformation
1Paweł Nowak, 1Roman Szewczyk, 2Robert Ugodziński, 3Piotr Bazydło
1Industrial Research Institute for Automation and Measurements (PIAP), 2Institute of Aviation, 3Research and Academic Computer Network (NASK)
pnowak@piap.pl, szewczyk@mchtr.pw.edu.pl, robertugo@gmail.com, piotr.bazydlo@nask.pl
Abstract: Paper presents problem of mesh selection optimization for most effective eddy current tomography forward transformation. Finite element methods of forward transformation have been previously reported. Although they are based on calculation for all test points, to ensure proper results reconstruction. Proceeding calculation of forward transformation with finite element method requires a lot of computation power and significant amount of time. In paper idea for characteristic point selection is presented. Simulation time can be limited, by selecting proper measurement points and interpolating the calculation results. Paper presents two methods of point selection. One is based on hybrid genetic algorithm and second one is discretely linear with randomly generated initial state. Furthermore comparison of computation time of utilized methods is presented, as well as calculation results. Presented results confirm possibility of effective result interpolation for more effective forward eddy current transformation.
47.
Comparison of Jiles-Atherton and bulk ferromagnetic hysteresis models for modelling the magnetic characteristics of advanced magnetic materials
Roman Szewczyk
Industrial Research Institute for Automation and Measurements (PIAP)
rszewczyk@piap.pl
Abstract: Paper presents the comparison of original Jiles-Atherton model and its modification, proposed by R. Venkataraman, called the bulk ferromagnetic hysteresis model. Due to the fact, that both models utilize the concept of anhysteretic magnetization, bulk magnetic hysteresis model was extended also to anisotropic materials. Comparison was carried out for isotropic Mn-Zn ferrite and for amorphous alloy with strong perpendicular anisotropy. The results indicate, that both models well reproduce the shape of magnetic hysteresis loops, however, some of model’s parameters vary significantly for presented models.
48.
Vectorization of the software for modelling the magnetostatic properties of thin layers using the method of moments
Roman Szewczyk
Industrial Research Institute for Automation and Measurements (PIAP)
rszewczyk@piap.pl
Abstract: Method of moments can be valuable alternative for finite elements method in the case of modelling the thin layer magnetostatic systems. However, “for”-loops based calculation of such system is may be very time consuming. Paper presents the method of vectorization of a code for method of the moments calculations. Radical decrease of computation time for larger systems was achieved due to the use of BLAS procedures.
49.
Characteristics of question of blind source separation using Moore-Penrose pseudoinversion for reconstruction of EEG signal
Szczepan Paszkiel
Institute of Control & Computer Engineering, Opole University of Technology
szczepanpaszkiel@o2.pl
Abstract: The paper presents question of blind source separation encountered by researchers aiming to determine location of generation electric activity in human brain as a source signal characteristic for given neuron fraction. To that end, Blind Signal Separation (BSS) technique with Moore-Penrose pseudoinversion was presented. The technique is useful for reconstruction of EEG signal. For the experimental purpose, sLORETA algorithm was also used to identify sources as a part of the inverse problem.
50.
Heat transfer in the thermo-anemometric flowmeter for biofuels
1Olena Bezvesilna, 2Marcin Kamiński, 3Andriі Ilchenko
1National Technical University of Ukraine “Kyiv Polytechnic Institute”, 2Industrial Research Institute for Automation and Measurements (PIAP), 3Zhytomyr State Technological University, Ukraine
olenabezvesilna@gmail.com, mkaminski@piap.pl, avi_7@rambler.ru
Abstract: This paper analyzes the heat transfer in the thermo-anemometric flowmeter tube for biofuels. The equation of heat transfer in the radial and axial directions of the tube anemometer was obtained. These equations take into account the properties of a “biological” component of the mixed fuel based on diesel fuel and gasoline upon measurement of their consumption by the thermo-anemometric flowmeter.
51.
Gravimeters of aviation gravimetric system: сlassification, comparative analysis, prospects
1Olena Bezvesilna, 2Marcin Kamiński
1National Technical University of Ukraine “Kyiv Polytechnic Institute”, 2Industrial Research Institute for Automation and Measurements (PIAP)
olenabezvesilna@gmail.com, mkaminski@piap.pl
Abstract: The article represents the results of a comparative analysis of existing gravimeters of aviation gravimetric system and identifies their strengths and weaknesses. It also represents and considers structures of the existing types of gravimeters that are used as elements of aviation gravimetric systems, and analyzes metrological characteristics of such sensitive elements. The paper describes modern advanced developments in the field of construction of innovative aviation gravimeters.
52.
Cross-sensor Calibration Procedure for Magnetometer and Inertial Units
Jakub Tomczyński, Tomasz Mańkowski, Piotr Kaczmarek
Poznań University of Technology
jakub.tomczynski@put.poznan.pl, tomasz.mankowski@put.poznan.pl, piotr.kaczmarek@put.poznan.pl
Abstract: Due to inertial and magnetic sensors imperfections, pre-processing is crucial in obtaining reliable orientation estimates. An easy to implement method of sensor calibration is presented, requiring little to none additional equipment. The method does not require a reference sensor and relies on world magnetic and gravity vectors constant relationship. In addition to standard individual calibration, a cross-sensor reference frame alignment is performed. Comparison with a high-grade MARG unit is also provided. Magnetic inclination stability, inertial and magnetic vector magnitude, and angular error are evaluated.
53.
Development of low cost speech system for patient suffer on the amyotrophic lateral sclerosis
Dominik Rybarczyk, Arkadiusz Kubacki
Poznań University of Technology
dominik.rybarczyk@put.poznan.pl, arkadiusz.j.kubacki@doctorate.put.poznan.pl
Abstract: The article describes design of low cost support speech system for patient suffer on the amyotrophic lateral sclerosis (ALS). The authors performed a literature overview concerning on the piezoelectric sensors, the ALS disease and construction systems supporting speech. The system was built with piezo sensor, which was triggered by the mimic face movements. The read signals is transmit to the extern application via USB interface, allowing patient to construction of words and phrases. The system has been tested, that allowed to choose the sensor optimal location on the head. Performed research is the starting point for building fully mobile speech system.
54.
Research on concentration levels depending on the color and blinking frequency of the marker using multiple EEG channel
Arkadiusz Kubacki, Łukasz Sawicki, Dominik Rybarczyk, Piotr Owczarek
Poznań University of Technology
arkadiusz.j.kubacki@doctorate.put.poznan.pl, lukasz.w.sawicki@doctorate.put.poznan.pl, dominik.rybarczyk@put.poznan.pl, piotr.owczarek@put.poznan.pl
Abstract: This paper presents the research on the concentration index. In the first section authors present general information on their brain-computer interfaces. They presented a system placement of the electrodes on a head and information about the human brain wave frequencies. In the next section the authors have presented the research methodology. They describe a method of carrying out three tests and equipment which they had used during the test. In the next chapter they presented the results of the tests. They have verified which electrodes must be taken into consideration when examining the concentration index. They showed how it affects the color and the frequency blinking marker change on the test results. The authors determined the best frequency for this type of research.
55.
Active LR Integrator Circuit for Drift-Free Fluxmeter
1Piotr Gazda, 1Michał Nowicki, 2Maciej Kachniarz, 1Maciek Szudarek, 1Roman Szewczyk
1Warsaw University of Technology, 2Industrial Research Institute for Automation and Measurements (PIAP)
p.gazda@mchtr.pw.edu.pl, m.nowicki@mchtr.pw.edu.pl, mkachniarz@piap.pl, mszudarek@gmail.com, r.szewczyk@mchtr.pw.edu.pl
Abstract: The following paper presents the idea of the active integrator circuit based on inductive and resistance components. Such design allows to eliminate the time drift of the circuit, which is undesired phenomenon resulting from ca-pacitive components utilized in classic constructions. This type of integrator circuit can be used to develop the drift-free fluxmeter for investigation of ferromagnetic hysteresis phenomenon in soft magnetic materials. The paper presents concept of the integrator circuit. The SPICE simulations was performed to validate the presented idea. Finally the developed solution was tested and obtained results confirm the correctness of the integrator circuits operation.
56.
Test stand for investigating of Giant Magneto-Impedance
1Piotr Gazda, 2Maciej Kachniarz, 1Maciej Szudarek
1Institute of Metrology and Biomedical Engineering, Warsaw University of Technology, 2Industrial Research Institute for Automation and Measurements (PIAP)
p.gazda@mchtr.pw.edu.pl, mkachniarz@piap.pl, mszudarek@gmail.com
Abstract: The paper presents a test stand that will be used in study of Giant Magneto-Impedance (GMI) effect. Measuring system is based on primary electric law and basic instruments used in electrical metrology. GMI phenomena occurs when a magnetic material is subjected to a DC axial magnetic field and a high frequency AC current. It results in large impedance changes (up to 300%) depending on the value of magnetic field. GMI effect sensors are widely used in weak magnetic field detection.Results of the first experiment that proves proper operation of the test stand are presented. Construction of a test bench is the first step to perform extensive study on this promising technology.
57.
Test stand for Matteucci Effect measurements
1Tomasz Charubin, 2Andrzej Juś
1Institute of Metrology and Biomedical Engineering, Warsaw University of Technology, 2Industrial Research Institute for Automation and Measurements (PIAP)
charubin@mchtr.pw.edu.pl, ajus@piap.pl
Abstract: In this paper test stand consisting of Joule annealing station, and magnetic field measurement stand for Matteucci effect investigation is described. The system al-lows for annealing of amorphous wire samples while applying torsion (twisting of the wire), and measuring the ME voltage response to alternating magnetic field applied to sample in chosen directions. This paper presents preliminary results of the measurements which will help modelling of ME in computer simulations.
58.
The use of remote ground sensing data for assessment of environmental and crop condition of the reclaimed land
1Lyudmyla Kuzmych, 2Igor Korobiichuk, 3Volodymyr Kvasnikov, 2Paweł Nowak
1National University of Water Management and Nature Resources Use, Ukraine, 2Industrial Research Institute for Automation and Measurements (PIAP), 3National Aviation University, Ukraine
l.v.kuzmych@nuwm.edu.ua, kiv_Igor@list.ru, kvp@nau.edu.ua, pnowak@piap.pl
Abstract: To solve the problems of environmental and crop monitoring, a Landsat remote sensing system is proposed, endowed with high spatial resolution of satellite images and the presence of spectral channels capturing the spectral characteristics of the studied objects in the visible and infrared spectra.
59.
Parameter-free on-line deep learning
Paweł Wawrzyński
Warsaw University of Technology
pwawrzynski@gmail.com
Abstract: In this paper the classic momentum algorithm for stochastic optimization is considered. A method is introduced that adjusts coefficients for this algorithm during its operation. The method does not depend on any preliminary knowledge of the optimization problem. In the experimental study, the method is applied to on-line learning in deep auto-encoders, and outperforms manually tuned coefficients. The method makes on-line learning a~fully parameter-free process and broadens the area of potential application of this technology.
60.
Low-effort place recognition with WiFi fingerprints using deep learning
Michał Nowicki, Jan Wietrzykowski
Poznań University of Technology
michalsminowicki@gmail.com, jan.wietrzykowski@put.poznan.pl
Abstract: Using WiFi signals for indoor localization is the main localization subsystem of existing personal indoor localization systems operating on mobile devices. The WiFi fingerprinting is also used for mobile robots as WiFi signals are usually available indoors and provide rough initial estimate or can be used in cooperation with other positioning systems. Currently, the best systems rely on manual data analysis, time-consuming parameter tuning and filtering to achieve a reliable and accurate localization system. In our work, we propose to use deep neural networks to significantly lower the work-force burden of careful system creation while still achieving satisfactory results. Assuming state-of-the-art hierarchical approach, we employ the DNN system for building/floor classification. We show that stacked autoencoders allow to efficiently reduce feature space in order to achieve a robust and precise system. Proposed architecture is verified on publicly available UJIIndoorLoc dataset and the results are compared with other solutions.
61.
Adopting the FAB-MAP algorithm for indoor localization with WiFi fingerprints
Jan Wietrzykowski, Michał Nowicki, Piotr Skrzypczyński
Poznań University of Technology
jan.wietrzykowski@cie.put.poznan.pl, michalsminowicki@gmail.com, piotr.skrzypczynski@put.poznan.pl
Abstract: Personal indoor localization is usually accomplished by fusing information from various sensors. A common choice is to use the WiFi adapter that provides information about Access Points that can be found in the vicinity. Unfortunately, state-of-the-art approaches to WiFi-based localization often employ very dense maps of the WiFi signal distribution, and require a time-consuming process of parameter selection. On the other hand, camera images are commonly used for visual place recognition, detecting whenever the user observes a scene similar to the one already recorded in a database. Visual place recognition algorithms can work with sparse databases of recorded scenes and are in general simple to parametrize. Therefore, we propose a WiFi-based global localization method employing the structure of the well-known FAB-MAP visual place recognition algorithm. Similarly to FAB-MAP our method uses Chow-Liu trees to estimate a joint probability distribution of re-observation of a place given a set of features extracted at places visited so far. However, we are the first who apply this idea to recorded WiFi scans instead of visual words. The method is evaluated on the UJIIndoorLoc dataset used in EvAAL competition, allowing fair comparison with other solutions.
62.
Review of 3D Objects Segmentation Methods
Maciej Wencka, Krzysztof Walas
Poznań University of Technology
maciej.wencka@gmail.com, krzysztof.walas@put.poznan.pl
Abstract: This paper presents a review of segmentation methods of basic shapes represented by polygonal meshes. For a fair algorithms comparison, common training data was used. In this work, 11 methods of 3D Mesh segmentation were tested using four different measures of segments similarity. Namely, Cut Discrepancy, Hamming Distance, Rand Index, Consistency Error were used. All measures mentioned above were characterised in the paper. The results of the comparisons provide means of understanding strengths and weaknesses of the tested algorithms and provide the foundation for the further developments of 3D Objects segmentation methods.
63.
Learning Fuzzy Cognitive Maps using Evolutionary Algorithm based on System Performance Indicators
1Katarzyna Poczęta, 1Łukasz Kubuś, 1Alexander Yastrebov, 2Elpiniki I. Papageorgiou
1Kielce University of Technology, 2Technological Educational Institute (T.E.I.) of Central Greece
k.piotrowska@tu.kielce.pl, l.kubus@tu.kielce.pl, a.jastriebow@tu.kielce.pl, epapageorgiou@teiste.gr
Abstract: Fuzzy cognitive map (FCM) is a soft computing technique for modeling decision support systems. Construction of the FCM model is based on the selection of concepts important for the analyzed problem and determining significant connections between them. Fuzzy cognitive map can be initialized based on expert knowledge or automatic constructed from data with the use of supervised or evolutionary learning algorithm. FCM models learned from data are much denser than those created by experts. This paper proposes a new evolutionary approach for fuzzy cognitive maps learning based on system performance indicators. The learning process has been carried out with the use of Elite Genetic Algorithm and Individually Directional Evolutionary Algorithm. The developed approach allows to receive FCM model more similar to the reference system than standard methods for fuzzy cognitive maps learning.
64.
Utilization of Deep Reinforcement Learning for Saccadic-based Object Visual Search
Tomasz Kornuta, Kamil Rocki
IBM Research, Almaden
tkornuta@gmail.com, kmrocki@us.ibm.com
Abstract: The paper focuses on the problem of learning saccades enabling visual object search. The developed system combines reinforcement learning with a neural network for learning to predict the possible outcomes of its actions. We validated the solution in three types of environment consisting of (pseudo)-randomly generated matrices of digits. The experimental verification is followed by the discussion regarding elements required by systems mimicking the fovea movement and possible further research directions.