INTED2013 Proceedings (ISBN: 978-84-616-2661-8)
Contents
1. 5 Movement parat 16 K of spring 32 f Maximum iterations per pani 64 K of dampe 128 a b Fig 5 a Number of Fourier components the user can use and b movement parameters the user can configure to do the experiment Function buttons Pause Continue Topological error 6 36 a b Fig 6 a The function buttons the user can use during map building and b the topological error of the mapping problem the user gets at the end of the process 4 IMPLEMENTATION DETAILS The implementation of the application has been made by employing a mathematical software tool that allowed us to make complex mathematical operations in a short period of time allowing us to develop a user interface that simplified the use of it With this purpose we decided to use MATLAB MATrix LABoratory mathematical software that offers an integrated elaboration environment with its own programing language M language 10 4 1 MATLAB MATLAB was created in 1970 by Cleve Moler It is a high level language and interactive environment that allows us to perform computationally intensive tasks MATLAB can be used in an extensive range of applications such as communications control systems design signal and image processing financial analysis and computational biology etc MATLAB includes vector and matrix operations that are primordial to solve scientific and engineering problems MATLAB incorporates a large number of g2. Image distance 30 cm r Axis X images 6 al Axis Y images 3 r Fourier Signature features Number ot Fourier 32 r components Movement parameters Maximum iterations per panicie 1000 Partide mass Time increment for movement K of spring K of damper Function buttons Start Pause SS a Stop Continue Topological error 6 36 Figure 1 Real Grid File Edit View Insert Tools Desktop Window Help ARATODEM 2BOA aD 220 a 09 024 025 40 1 180 67 10 22 26 2 42 180 i 6 11 2 027 3 es 012 021 025 6 7 4 Oo 12 20 029 036 das 14 819 930 046 15 Sis 031 934 947 617 32 Fig 3 Appearance of the graphical interface of the application Main features Main features Main features Database Database Database Laboratory g Mathod Method Method ing S Image distance Image distance Image distance K H Axis X images Axis X images Axis X images pi Axis Y images Axis Y images Axis Y images s 9 a c Fig 4 a Databases the user can choose b map building methods the user can use and c number of images in each axis the user can choose to do the experiment 3079 Movement parameters Maximum iterations per particle Fourier Signature features Particle mass Number of Fourier a S rate components ime increment tor movement
3. In this subject the students learn techniques for map building and localization of mobile robots using the appearance based information extracted from the images captured by the camera the robot carries on it Thanks this tool the students can fully understand the appearance based approach in robotics mapping and can experience the localization process with real data The main features of the tool are Some databases with panoramic images both grey scale and colour of several real environments are included With these images the student can test the algorithms they have learned to build the map We have implemented a method to compress the information of the images based in the discrete Fourier Transform of images We have implemented an algorithm to build a map using an approach based on a set of forces created by some virtual springs among the images the robot has captured along the environment Students can test how some parameters affect the final map this algorithm computes The student can test how the degree of compression of the visual information affects to the final map The tool is fully interactive It shows the map building process step by step graphically and at the end it shows the layout of the final map and it gives a measure of the error As all the databases included have been captured in real environments under realistic lighting conditions the tool allows the students to face a real problem and experi
4. can work with the images as matrices Another important function within the library of MATLAB functions correspond to fft m used to obtain the FFT Fast Fourier Transform of a sequence of numbers To filter the images we use homofilter m lt has been developed by us as it has not been found in the library of MATLAB functions It returns a filtered image using a Homomorphic filter With regards to the movement of poses in the map we have used the function random m to give a random initial location at each pose of the map In addition we have used a function that allows us to obtain the topological error between the real map and the map obtained with our methods The function used is procrustes m u procrustes X Y and determines a linear transformation translation reflection orthogonal rotation and scaling of the points in matrix Y to best conform them to the points in matrix X The function returns the minimized value of this dissimilarity measure in u To finish we have used two functions to make possible the interaction between the graphical interface and the application code get and set The first one obtains the status of any of the properties of an object in the GUI The second one allows us modify any property of a GUI object Thus with both commands the interaction with the user is performed 9 CONCLUSIONS This work presents a software tool we have designed to be used in a robotics and computer vision subject in a PhD degree
5. of map creation using visual information is enormous and new algorithms are published continuously This way both computer vision and robotics constitute two fields in engineering which are in a continuous development and that permit solving with robustness tasks that imply the autonomous movement of a robot in a real scenery 1 In a subject of our PhD degree in the Miguel Hernandez University Spain there is an optional subject where these topics are studied Students learn how the information extracted from a set of images captured by the robot can be used during the map building and localization process The students study some classical approaches to solve these problems and the practical sessions of this subject allow the students to put into practice the theoretical concepts studied in the classroom Traditionally these practical sessions have been developed in the robotics laboratory The students are proposed some tasks they have to solve making use of a robot and the visual information it captures with a camera it carries on it The students have to implement the necessary algorithms to solve them The main problem of these sessions is the fact that the number of robots and cameras available in the laboratory to develop and test the algorithms is limited and so the available time in the laboratory Proceedings of INTED2013 Conference ISBN 978 84 616 2661 8 4th 6th March 2013 Valencia Spain 3074 Our objective is to overcome these dif
6. the collection of conference papers Searching an index is much faster than searching all the text in the documents To search the INTED2013 Proceedings index 1 Open the Search PDF pane through the menu Edit gt Advanced Search or click in the PDF bookmark titled SEARCH PAPERS CONTENT 2 The INTED2013_index pdx should be the currently selected index in the Search window if the index is not listed click Add locate the index file pdx on the CD and then click Open 3 Type the search text click Search button and then proceed with your query For Acrobat 9 1 In the Edit menu choose Search You may receive a message from Acrobat asking if it is safe to load the Catalog Index Click Load 2 A new window will appear with search options Enter your search terms and proceed with your search as usual For Acrobat 8 1 Open the Search window type the words you want to find and then click Use Advanced Search Options near the bottom of the window 2 For Look In choose Select Index 3 In the Index Selection dialog box select an index if the one you want to search is available or click Add and then locate and select the index to be searched and click Open Repeat as needed until all the indexes you want to search are selected 4 Click OK to close the Index Selection dialog box and then choose Currently Selected Indexes on the Look In pop up menu 5 Proceed with your search as usual selecting other
7. 13 International Technology Education and Development Conference INTED2013 COMMITTEE AND ADVISORY BOARD Agust n L pez Alan McGowan Alessio Artuffo Alexander Gromoff Alma Beatriz Rivera A guilera Alona Senishch Chmilewsky Alvaro Torres Amparo Gir s Anthony Edwards Antonio Garc a Canan Karababa Caroline Rizza Catherine Marinagi Chelo Gonz lez Cristina Lozano Darryl Corey David Mart Dianne Chambers Eladio Duque Euclid Keramopoulos Eva Catherine Hillemann Evi Papaioannou Faith Gudo Gilda Rosa Bola os Harm Tillema Haydar Oztas Heidi Flavian Hulya Gorur Atabas Ignacio Ballester Ignacio Candel loanna Rigina Karariga Ismael Serrano Iv n Mart nez Ivan Traina Jae Dong Han Javier Domenech Javier Mart Jayant Ghiara Jesudasan Fredrick Thomas Joanna Lees John Murray Jose F Cabeza Jose Luis Bernat Josephat Oroma Judith Szerdahelyi Kartikay Saini Katja Zaki Ketevan Kupatadze SPAIN UNITED STATES ITALY RUSSIAN FEDERATION MEXICO FINLAND GUATEMALA SPAIN UNITED KINGDOM SPAIN TURKEY ITALY GREECE SPAIN SPAIN UNITED STATES SPAIN AUSTRALIA SPAIN GREECE AUSTRIA GREECE BOTSWANA MEXICO NETHERLANDS TURKEY ISRAEL TURKEY SPAIN SPAIN SAUDI ARABIA SPAIN SPAIN ITALY CANADA SPAIN SPAIN UNITED STATES OMAN FRANCE UNITED KINGDOM SPAIN SPAIN TANZANIA UNITED STATES INDIA GERMANY UNITED STATES Kiruthika Ragupathi Koen Slootmaeckers Krista Lussier Krist
8. IN PED 201 1th International Technology Education and Development Conference Valencia Spain 4th 6th of March 2013 CONFERENCE PROCEEDINGS Ww mtedZ013 org VJ DEB 7th International Technology Education and Development Conference Valencia Spain 4th 6th of March 2013 p gt A 1 gt b h o PO A F ud y N gt AAA y PQ PAN N 65 0 rm rm a C B C f 4 A a h AS X N L L N L L E L L www inted2013 org Published by International Association of Technology Education and Development IATED www iated org INTED2013 Proceedings 7th International Technology Education and Development Conference March 4th 6th 2013 Valencia Spain Edited by L G mez Chova A L pez Martinez I Candel Torres International Association of Technology Education and Development ATED ISBN 978 84 616 2661 8 Deposito Legal V 424 2013 Book cover designed by J L Bernat All rights reserved Copyright 2013 ATED The papers published in these proceedings reflect the views only of the authors The publisher cannot be held responsible for the validity or use of the information therein contained Some conference presentations may
9. anoramic images it is necessary to take into account the fact that appearance is influenced both by the position and shape of the objects and the illumination conditions The appearance of an object in an image can vary strongly depending on the illumination systems and the lighting level of the scene whether it is artificial or natural and the part of the day when the images were captured It is therefore necessary to implement a mechanism that allows us to work independently of the lighting conditions of the environment The application of several kinds of filters to the images offers us invariance with respect to the illumination of the scene in object recognition tasks In these applications the student may choose to apply a homomorfic filtering It has shown a good performance in removing the contribution of changing lighting conditions in the scenes 8 After this process each image is represented by a vector d iE4 n where n is the number of images of the environment to map The size of each vector is sxr where s is the number of rows of the panoramic images and r is the number of Fourier components retained to build the descriptor r is a configurable parameter The larger is r the most information is retained from the scenes but the higher is the computational cost 2 3 Computing the topological or geometrical relationships between descriptors In this application we propose the students to use a mass spring damper model to bu
10. configuration of the algorithms Some databases have been added to this tool Each of them is composed of a set of real indoor images captured in a real environment under realistic lighting conditions Thanks to this software the students can learn how this family of methods works and the importance of a good choice of the parameters to get accurate results Keywords Computer vision mobile robots appearance based methods educational software higher education 1 INTRODUCTION VIRTUAL LABORATORIES IN MOBILE ROBOTS AND COMPUTER VISION Nowadays the presence of robots around us is continuously increasing both in our houses and in the factories They have appeared in such environments to carry out some tasks that are unpleasant for us However it is necessary these robots have a high degree of autonomy to be able to develop correctly in the environment where they have been designed to work Very often it is necessary that the mobile robot is able to know where it is located and to calculate how to arrive to the target points to complete the task it is developing With this aim the robot must keep an internal representation of the environment where it evolves We name this representation a map of the environment The use of computer vision to create a representation of the environment is very extended due to the good relationship quantity of information cost that the cameras offer The research developed during the last years in the topic
11. des a parameter u 0 that is a measure of the shape correspondence between the sets of points The lower is u the more similar are the two sets We name this parameter u Topological error 9 3 DESCRIPTION OF THE TOOL In this section we present a detailed description of the performance of our application Our general philosophy while designing this tool was the simplicity of use We try to guide the user during the whole process and take into account that it is possible that the user may not be an expert in the autonomous robot field The final objective of the tool is to provide the students with a set of real databases of omnidirectional images to understand and test the feasibility of an appearance based topological mapping approach using a spring mass damper system to create a map of the environment 3 1 User s manual We use MATLAB 10 to develop the application It is necessary the student run first MATLAB in his computer because the application must be launched from this program When it starts running a graphical interface appears All the necessary actions can be carried out through this interface The appearance of the graphical interface is shown in fig 3 In general this figure shows all the necessary options to build the topological map a bird s eye view of the real map distribution of the images the final topological map obtained with the algorithm and the topological error The topological map is updated step by s
12. e of the map obtained Once the user has chosen that parameters it is necessary to adjust the time increment for the movement of the system amp and the value of the constant k of the spring and damper to simulate the performance of each spring in the forces system fig 5 b Once all the parameters have been selected the user will press the Start button to begin the mapping process During the process the students can appreciate an interactive evolution of the spring mass damper topological mapping algorithm performance The student can see the topological map blue points in fig 3 in each step of the process When the student has chosen the incremental method the process gradually adds new positions on the map and stabilizes them In the opposite case batch mapping all poses are added in the first moment in random positions and then the system is allowed to tend to balance In addition the topological error in the map building is computed and shown at the end of the process fig 6b Finally the user can pause the process by 3078 pressing the Pause button and later continue with it by pressing the Continue button or reset the process by clicking on the Stop button Spring_Mass A 0 gt 27 E APPEARANCE BASED MAPPING WITH PANORAMIC IMAGES AND SPRING MASS DAMPER Main features Database Laboratory ne Method Incremental pd
13. ence the troubles that would outcome in a real application This way we provide students with a tool that allows them to freely test and improve the algorithms they learn in the classroom The tool has demonstrated to be useful for the students to understand the mapping algorithms and other basic concepts in mobile robotics and computer vision We expect that thanks to this tool the students are able to design and develop new algorithms in map building and localization 3081 ACKNOWLEDGEMENTS This work has been supported by the Spanish government through the project DPI2010 15308 Exploraci n Integrada de Entornos Mediante Robots Cooperativos para la Creaci n de Mapas 3D Visuales y Topol gicos que Puedan ser Usados en Navegaci n con 6 Grados de Libertad REFERENCES 1 2 3 4 2 6 7 8 9 10 Siegwart R 4 Nourbakhsh I R Introduction to autonomous mobile robots The MIT Press 2004 ISBN 0 262 19502 X Pay L Reinoso O Torres F Puente S T A Web based Platform for Remote Interaction with Mobile Robots in Higher Education International Journal of Engineering Education 2011 ISSN 0949 149X 27 Moller R Vardy A Kreft S Ruwisch S Visual homing in environments with anisotropic landmark distribution Autonomous Robots 2007 ISBN 0929 5593 Goecke R Asthana A Pettersson N Petersson L Visual Vehicle Egomotion Estimation using the Fourier Mellin Trans
14. ficulties and also to make an adaptation to the new philosophy of the European Space for Higher Education that must be centred in the learning and autonomy of the student it should be a less rigid system with regards to the lectures and timetables With this aim we are encouraged to use some new methodologies to improve the autonomy of the students 2 We have designed and implemented a software tool that facilitates the students to reach the objectives of the subject This software constitutes a link between the data collected by the real robots and the task the student has to carry out This way after an initial presence practical session where the teacher exposes the fundamentals of the robots and the cameras the student can develop the rest of the sessions in his house with a flexible timetable making use of the software we have designed When any problem or doubt arises students can get in touch with the lecturer through a virtual tutoring system The remainder of the paper is structured as follows section 2 presents the principles the students study in the theoretical classes In section 3 we detail how the software tool works Section 4 presents some implementation details and at last in section 4 we present the conclusions of the work 2 APPEARANCE BASED METHODS IN ROBOTICS APPLICATION TO MAP BUILDING In this section we present a summary of the theoretical background that students learn about appearance based map building befo
15. form Proc of IEEE Intelligent Vehicles Symposium 2007 ISBN 1 4244 1067 3 Fernandez L Paya L Reinoso O Amor s F Appearance based Visual Odometry with Omnidirectional Images A Practical Application to Topological Mapping Proc of the International Conf on Informatics in Control Automation and Robotics 2011 ISBN 978 989 8425 75 1 Fernandez L Paya L Valiente D Gil A Reinoso O Monte Carlo Localization using the Global Appearance of Omnidirectional Images Algorithm Optimization to Large Indoor Environments Proc of the International Conf on Informatics in Control Automation and Robotics 2012 ISBN 978 989 8565 22 8 Menegatti E Maeda T Ishiguro H Image Based Memory for Robot Navigation Using Properties of Omnidirectional Images Proc of Robotics and Autonomous Systems 2004 Fernandez L Paya L Reinoso O Gil A Julia M Robust Methods for Robot Localization Under Changing Illumination Condittions Comparison of Different Filtering Techniques International Conference on Agents and Artificial Intelligence 2010 ISBN 978 989 674 021 4 Paya L Fernandez L Gil A Reinoso O Map Building and Monte Carlo Localization Using Global Appearance of Omnidirectional Images Sensors 2010 ISSN 1424 8220 10 http www mathworks com products matlab 3082
16. gn INTED2013 International Technology Education and Development Conference VIRTUAL SESSIONS Barriers to Learning Digital Divide amp Diversity issues in Education Collaborative and Problem based Learning Competence Evaluation Computer Supported Collaborative Work Curriculum Design and Innovation E content Management and Development E learning M learning and Blended Learning Educational Software and Serious Games Enhancing learning and the undergraduate experience Evaluation and Assessment of Student Learning Experiences in Education Experiences in Research in Education Globalization and Ethical issues in Education International Projects Learning and Teaching Methodologies Learning Experiences in Primary and Secondary School Lifelong Learning Links between Education and Research New challenges for the Higher Education Area New projects and innovations New Trends in the European Higher Education Area Pedagogical amp Didactical Innovations Quality assurance in Education Research Methodologies Research on Technology in Education Student Support in Education Technological Issues in Education Technology Enhanced Learning Tutoring and Coaching University Industry Collaboration Virtual Universities and Distance Education INTED2013 International Technology Education and Development Conference ABOUT INTED2013 Proceedings CD HTML Interface Navigating with the Web browser This CD includes all presented papers a
17. ild a topological map of the environment where the images layout matches up with the original layout Each image acts as a particle P iE4 n whose mass is m Each pair of particles P and P are joined with a spring S with elastic constant k and a damper with damping constant Ki If the value of the elastic constants is proportional to the distance between image descriptors then when this particle system evolves freely to the balance it is expected that the final layout of the particles is similar to the real layout where images were captured In our experiments we make the natural length of each spring Lio Di where D is the Euclidean distance between the descriptors of images and Taking into account the resulting force on each particle the position of this particle can be updated by applying iteratively the following expressions a F J v t AT AT Flt AT 5 0 5 4 AT Where a V f are respectively the acceleration velocity and position of the particle i whose mass is m The parameter AT that appears in these equations has a significant relevance in the necessary time to tend to balance and in the accuracy of the final particle distribution It is desirable this parameter to take a relatively high value along the initial steps when the particles may be far from balance and they should move quickly However as the system tends to balance this parameter should be given a lower val
18. ine Helen Falgren Laura loana Leon Linda Tribuzio Lorena L pez Louise Emanuel Luis G mez Chova MP Jes s Suesta Maria Perifanou Maria Porcel Maria Renata Duran Maria Teresa Gastardo Mark Carthew Mary Ellis Michael Cameron Michael Steinmann Michael Zanko Moira Hunter M nica Fern ndez Naser Zaeri Norma Barrachina Olga Teruel Omar Majid Pamela King Patricia Arnold Patrick Giddy Paulo Cunha Philippos Pouyioutas Regis Kawecki Ricardo Santa Rodolfo Matos Roger Hill Rudolf Maly Sarah List Sergio P rez Simon Collin Soane Mohapi Stephane Jacquemart Sudha Goyal Svetlana Titova Tatjana Vasiljeva Thelma de Jager Thor O Olsen Ulrich Borgolte Victor Fester Xavier Lefranc SINGAPORE BELGIUM CANADA ICELAND ROMANIA UNITED STATES SPAIN UNITED KINGDOM SPAIN SPAIN GREECE SPAIN BRAZIL GREECE AUSTRALIA SINGAPORE NEW ZEALAND NETHERLANDS AUSTRALIA FRANCE SPAIN KUWAIT SPAIN SPAIN MALAYSIA UNITED STATES GERMANY SOUTH AFRICA PORTUGAL CYPRUS FRANCE SAUDI ARABIA PORTUGAL UNITED STATES GERMANY AUSTRALIA SPAIN CANADA SOUTH AFRICA LUXEMBOURG INDIA RUSSIAN FEDERATION LATVIA SOUTH AFRICA NORWAY GERMANY NEW ZEALAND FRANCE INTED2013 International Technology Education and Development Conference CONFERENCE SESSIONS ORAL SESSIONS 4th March 2013 Mobile Learning Joint Educational and Research projects Collaborative and Problem based Learning University Industry Collaboratio
19. n 1 Technological Issues in Foreign Languages Education 1 Evaluation and Assessment of Student Learning Impact of Education on Development Research in Technology in Education International Projects Collaborative Experiences in Web Technologies University Industry Collaboration Workplace Learning Technological Issues in Foreign Languages Education 2 Virtual Universities Diversity and Ethical issues in Education Educational Software and Serious Games Curriculum Design Strategies Principles and Challenges Collaborative and Virtual Learning Environments University Industry Collaboration 2 Experiences in Foreign Languages Education ICT Skills and Competencies among Teachers Experiences in Business amp Adm Mgmt Education Blended Learning New Experiences for Curriculum Design Learning and Teaching Methodologies 1 Adult and Vocational Education Experiences in Foreign Languages Education Research Pre service Teacher Experiences Technological Issues in Business Adm Mgmt Education POSTER SESSIONS 4th March 2013 Poster Session 1 Research and General Issues in Education Poster Session 2 Pedagogical Innovations and New Trends in Education INTED2013 International Technology Education and Development Conference ORAL SESSIONS 5th March 2013 Web 2 0 and Social Networking 1 New Trends in the Higher Education Area Quality assurance in Education 1 Research in Education Experiences in Engineeri
20. ng Education Experiences in Biosciences Education amp Research Barriers to Learning Web 2 0 and Social Networking 2 New projects and innovations in Education 1 Quality assurance in Education 2 Virtual Universities and Distance Education Research in Engineering Education Pedagogical Innovations and Experiences in Health Enhancing learning and the undergraduate experience 1 Social media amp Social networks in Education New projects and innovations in Education 2 New challenges for the Higher Education Area 1 Primary amp Secondary Education Technological Issues in Engineering Education Technology Enhanced Learning in Health Education Enhancing learning and the undergraduate experience 2 Technology Enhanced Learning Learning and Teaching Methodologies 2 New challenges for the Higher Education Area 2 Learning Experiences in Primary and Secondary Education Language Learning Innovations 1 Technological Issues in Education Pedagogical amp Didactical Innovations in Architecture Education E learning General Issues Organizational legal and financial issues New challenges for the Higher Education Area 3 Educational Research Language Learning Innovations 2 E content Management and Development Education and Globalization POSTER SESSIONS 5th March 201312 Poster Session 1 Technology in Education and University Industry Collaboration Poster Session 2 Experiences in Education and Curriculum Desi
21. not be available for publication INTED2013 International Technology Education and Development Conference WELCOME INTRODUCTION Dear INTED2013 participants We are delighted to welcome you all to this 7th International Technology Education and Development Conference Today we are living in a technology based society where education and innovation are the key to the world s development and progress For this reason it is essential to be updated with new teaching and learning methodologies and explore new horizons in educational cooperation INTED2013 welcomes experts from more than 65 different countries This will be the best occasion to open your minds to other experiences and share best practices in education This forum will also provide an ideal networking platform to make new international contacts and partnerships We hope you benefit from the conference programme its interactive sessions and social activities where you will be able to meet other educators researchers and technologists from all continents and cultures Additionally we invite you to discover and enjoy the beautiful city of Valencia Do not miss the opportunity to walk around and visit its impressive architecture historical buildings green areas and lovely beaches Thank you very much for participating in INTED2013 and for contributing to this inspiring international forum We hope you enjoy your time with us INTED2013 Organising Committee INTED20
22. options you want to apply and click Search For Acrobat 7 and earlier 1 Inthe Edit menu choose Full Text Search 2 A new window will appear with search options Enter your search terms and proceed with your search as usual A NEW RESOURCE IN THE TEACHING OF A COMPUTER VISION AND ROBOTICS SUBJECT L Paya L Fernanez O Reinoso F Amoros and L M Jimenez Departamento de Ingenieria de Sistemas y Automatica Miguel Hernandez University SPAIN lpaya umh es I fernandez umh es Abstract In the field of mobile robotics the task of building a map of the environment where the robot moves is a very important step during the design of any application that requires the robot to move autonomously The most important features concerned in map building are deeply studied in advanced subjects about robotics As this topic is currently being developed by many researchers and new methods are being implemented it is necessary to provide students with a tool that allows them to understand how these methods work and how they can configure them to optimize the map built In this work we present a tool we have developed to be used in a computer vision and robotics subject in a PhD degree Using this tool the student will be able to test and understand many map building concepts in robotics using visual information and appearance based methods This tool offers an easy and intuitive graphical interface that guides the student though the
23. points situated on a regular grid We have transformed these omnidirectional images into panoramic images The main features of these databases are shown in table 1 Table 1 Relevant parameters of the image sets used in the mapping experiments Resolution Environment Environment Size images pais _ 126 Laboratory 563255 2 4x4 5 m Kitchen 11 8 58258 01 1x07m 8 25810 586258 01 3409 3075 2 2 Computing the appearance descriptor of each image In the appearance base approaches all the necessary comparisons between images to build the map are implemented using the information of the whole images without extracting any kind of landmark However working with the entire images presents a very high computational cost and also large memory requirements As an intermediate step in the appearance based approaches it is necessary to build a descriptor that retains the most relevant information of the images in a lower dimensionality Fourier representations constitute a traditional way of describing the images when working with panoramic scenes 4 5 Among these representations the Fourier Signature presents two interesting properties The most relevant information concentrates in the low frequency components of each row and it presents rotational invariance what is very interesting in robot localization applications 6 Further information can be found in 7 When we work with the appearance of p
24. raphic functions for visualizing data and also includes a large number of mathematical functions such as those to solve linear algebra problems in an optimal manner which enable us to perform Fourier analysis in a reasonable computational time 4 2 GUI We can design a graphical user interface in MATLAB through the tool GUIDE Graphical User Interface Development Environment With this tool we can add push buttons radio buttons selection lists dropdowns and sliders as well as ActiveX controls and plots The creation of a graphical user interface by GUIDE in MATLAB gives as a result two files matlabguide fig and matlabguide m The first one contains the graphical part and allows us to make the design of the appearance of the user interface The second one is the program code including the functions and command used by the application When we add a new push button or other object in the graphics window we have to write the code associated to that object in the suitable part of matlabguide m On the other hand we can also request the status of the properties of some objects in the graphics windows at each instant or change these properties if we need it 3080 4 3 Main functions The objective of this section is to outline the most interesting MATLAB mathematical functions we have used in our code for the application One important function is imread m which allows us to get an MxNx3 matrix of each image in the database and thanks to it we
25. re starting to use this software tool From our point of view the map building process consists of several phases which will be presented along this section 1 Capturing the images of the databases In a real application a robot with a camera mounted on it would travel along the environment to map It would capture some images along this environment and this set of images would constitute the basic information to build a map or representation of the environment 2 Computing an appearance descriptor per scene that contains the most relevant information of the image 3 Extracting the topological or geometrical relationships between descriptors It must be done by comparing the appearance descriptors 4 Comparing the layout of the computed map with respect to the geometry of the original map This final step is only necessary when we are testing a new algorithm to know if it leads to good results At the end of the process the set of image descriptors and the topological or geometrical relationships between them will constitute the map of the environment In the next subsections we develop these steps 2 1 Images databases To test the map building algorithms we have captured several databases with images from different environments The image databases we have used to carry out the experiments belong to the Technique Faculty of Bielefeld University 3 These databases consist in a set of omnidirectional images captured from a set of
26. roach must be implemented In this approach the map can be built online as the robot is going though the environment When we have a balanced set of particles and a new particle image arrives first we allow this new particle to tend to balance while maintaining the position of the other particles still and when it arrives to balance we allow the whole system to tend to balance fig 2 Figure 1 Example of some intermediate steps during the batch mapping process In step 1 the particles are situated at random positions and as the process advances and particles tend to balance their distribution tends to be similar to the real one Figure 2 Example of some intermediate steps during the incremental mapping In step 1 the new particles are added gradually to a previously balanced map in each step 3077 2 4 Shape difference between the original and the resulting map layout Once the map is built to test the accuracy of the algorithm we must evaluate how similar are the computed layout and the real layout of the captures As the only information we have to build the map is the distance between Fourier descriptors the resulting map is expected to have a similar shape comparing to the original grid but with a scale factor a rotation and a possible reflection Our tool makes use of a method based on the Procrustes analysis 7 This analysis allows us to compare two sets of points while removing these effects and as result it provi
27. t INTED2013 conference It has been formatted similarly to the conference Web site in order to keep a familiar environment and to provide access to the papers trough your default Web browser open the file named INTED2013 htm1 An Author Index a Session Index and the Technical Program are included in HTML format on this disk to aid you in finding conference papers Using these HTML files as a starting point you can access other useful information related to the conference The links in the Session List jump to the corresponding location in the Technical Program The links in the Technical Program and the Author Index open the selected paper in a new window These links are located on the titles of the papers and the Technical Program or Author Index window remains open Full Text Search Searching INTED2013 index file of cataloged PDFs If you have Adobe Acrobat Reader version 6 or later www adobe com you can perform a full text search for terms found in INTED2013 proceedings papers Important To search the PDF index you must open Acrobat as a stand alone application not within your web browser i e you should open directly the file INTED2013 pdf in the CD with your Adobe Acrobat or Acrobat Reader application This PDF file is attached to an Adobe PDF index that allows text search in all PDF papers by using the Acrobat search tool not the same as the find tool The full text index is an alphabetized list of all the words used in
28. tep so that student can understand the method From now on we describe the steps that must be carried out during a common sequence of use of the tool and the options it provides to the students The first step consists in configuring the parameters of the database that the user wants to use to build the map This step allows him her to choose from five different images databases table 1 Also he she can choose between the batch and the incremental options to build the map Finally it permits choosing between different distributions within a image database To compute the distribution of the original map the student must choose the number of images in each axis of the system and the distance between them fig 4 In the second step of the process the student has to choose the number of Fourier components 7 used to compute the Fourier Signature of each panoramic image in the database With this parameter the user will be able to understand the importance of the compression level in the final result of the experiment and the necessity to find an appropriate number of Fourier components for the mapping process to work appropriately fig 5 a Finally the student has to choose the parameters involved in the spring mass damper system First the user has to choose the maximum iterations per particle 5 and the mass of each particle m and test the importance of this parameter both in the time necessary to build the map and in the shap
29. ue as the particles are expected to be nearer their final positions We 3076 propose to set a maximum number of steps s and we make AT to change along the process 5 is the current step in this process S tot are A good choice of the parameters amp k and K is important for the convergence of the system and to arrive to a good distribution of the balanced particles The elastic constants are important because they have influence in the speed of the particles while they tend to balance On the other hand the damping constants have an important role in the convergence of the mapping process Thanks to them the change in the positions of the particles between two consecutive iterations is not a sudden process and the system tends to balance gradually At last the process followed to build the map depends strongly on how the images arrive along the process When all the images are available before the mapping process begins and we have no information about the order they have arrived in a batch mapping approach must be used In this approach all the particles are given random initial positions and then we allow the system to tend to balance to form the final map fig 1 We consider the system has arrived to balance when the sum of movements of all particles is under a threshold On the other hand if the images are captured gradually and the mapping process must start before the whole set of images is available an incremental app
Download Pdf Manuals
Related Search