intelligent systems group (isg) - infotech oulu · 28 infotech oulu annual report 2006 intelligent...

28 INFOTECH OULU Annual Report 2006

INTELLIGENT SYSTEMS GROUP (ISG)

Professor Juha Röning, Professor Jukka Riekki and Professor Tapio Seppänen,

Computer Engineering Laboratory, Department of Electrical and Information Engineering,

University of Oulu

[email protected], [email protected], [email protected]

http://www.ee.oulu.fi/research/isg

Background and Mission

The Intelligent Systems Group’s (ISG) mission is to carryout long-term research on novel technologies and applica-tions of intelligent systems. The main objective is to de-velop enhanced adaptivity and context-awareness for smartenvironments. The research specifically focuses on the crea-tion of dynamic models that enable monitoring, diagnos-tics, prediction and control of target systems (living andartificial) or operating environments. It is our aim to makethe environment adapt to the users, instead of making theusers adapt to an inflexible environment. We believe that,by creating these novel components for smart environments,important enabling functionality will emerge that will mul-tiply the versatility and applicability of such living envi-ronments.

We see behaviour modelling as a major challenge in devel-oping truly intelligent and proactive environments. Humanusers of smart environments often behave in such a com-plex manner that it is hard to predefine and preprogram allof their behavioural patterns in the software. Models of userbehaviour are required that are able to grasp the user’s con-text at any moment and to enable adaptation of the func-tionality of the intelligent environment to the situation athand. Further, it is essential to model the behaviour of thedevices controlled by the intelligent environment, whichenables adaptation to environmental changes without re-programming. Systems should eventually learn and adaptautomatically through these models, and thus perform theirduties effectively.

Our research group combines a variety of key skills andtechnologies to attain this goal. We have experience of thefollowing key technologies: system architectures and im-plementation of context-aware systems; modelling and rec-ognition of contexts from sensor signals; data miningalgorithms; nomadic learning robots; embedded systemstechnologies; software security; and smart environmentimplementations. The key application areas are: smart liv-ing environments of homes and institutes; industrial auto-mation; mobile robots; context-aware mobile devices; andwellness and medical applications. Each of these domainspossesses special characteristics but, from the point of viewof developing algorithms for an intelligent system, they alsopossess remarkable similarities. They all produce a multi-tude of signals that represent the status of the system. Thetarget system behaviour should be modelled and recognizedbased on the signals. The application service should thenact accordingly. The availability of several application do-mains yields many advantages: a solution to a special prob-lem in one domain may offer added-value functionality in

some other domain; our solutions are deployed by many ofour client industries; solutions to a wide range of real-worldproblems define a credible and versatile tool-box that has amajor impact on our development-oriented subcontractualprojects.

The group co-operates with many international and domes-tic partners. In applied research, the group is active in Eu-ropean projects, and several joint projects funded by theFinnish Funding Agency for Technology and Innovation(Tekes) and by industry.

The group and its members are active in the scientific com-munity. For example, Prof. Juha Röning served as Co-chairon the program committee of Intelligent Robots and Com-puter Vision XXIV: Algorithms, Techniques, and ActiveVision (1-4 October 2006, Boston, Massachusetts, USA).In the software security area The European Intensive Pro-gramme on Information Security Management and Tech-nology 6th Winter School was arranged in Oulu 4-12 April,2006 and The Second Crisis Management Workshop(CRM2006) in Rovaniemi. Finland, September 28, 2006.Prof. Riekki arranged a co-operative seminar in Sendai,Japan in Spring 2006. Also several members of the groupwere on the committees of international conferences. Thegroup’s expertise is recognized; for example, one surveyfor the Finnish Ministry of Transport and Communicationswas prepared as well as several invited talks and lectureswere given.

The third To be or Well be seminar on wellness was ar-ranged in Oulu. Prof. Seppänen was one of the main organ-izers. Almost 200 visitors from all over Finland participatedthe seminar. Prof. Seppänen was a leading figure in estab-lishing the WellTech Oulu institute for wellness and medi-cal technology in the University of Oulu. The new institutecoordinates the research and education in this area, as wellas participates in various activities in the operating envi-ronment. Prof. Seppänen was also the leading figure in theestablishment of the Oulu School of Biomedical Engineer-ing, an umbrella organization of wellness and medical tech-nology education at the University of Oulu and the OuluUniversity of Applied Sciences.

We have talented members in the team. The Master’s thesis“Exploiting Communication Patterns in Complex Informa-tion Networks” by Jani Kenttälä received two awards, theTelecom Scholarship 2005 (best telecom related thesis inFinland published in 2005) and the Finnish Association forMathematicians, Physicists and Computer Scientists(SMFL) prize for the best pro gradu thesis in 2005.

Interest in our research has been strong. For example this

INFOTECH OULU Annual Report 2006 29

year the EU Press delegation from Japan, Korea and Ma-laysia and the Finnish Minister of Education and Sciencevisited our research group. During the visits, robotics re-search was demonstrated.

The activities of the ISG are led by professors Juha Röning(Director), Jukka Riekki and Tapio Seppänen.

Scientific Progress

In 2006 the research at ISG concentrated on prototypingsmart environments, mobile and context-aware systems, datamining methods, signal analysis and secure programming.These were applied in context-aware mobile systems, in-telligent service robots, steel plant and spot welding proc-ess quality control, and analysis of biomedical ECG andEEG signals.

Research on prototyping: from a smart envi-

ronment towards remote distributed intelli-

gence

Verification of the developed methods and models in pro-totypes will be an important part of the research. To sup-port this activity, we will develop software and hardwarearchitectures for smart environments. In addition to verifi-cation, prototypes speed up the commercialization of theresearch results. In prototyping, we have set and tackledthe following objectives:

Developing software architectures for a smart environment.

The aim is to create a software architecture supporting thedevelopment of context-aware applications for a smart en-vironment. We have developed a Property Service Archi-tecture, a modular and scalable software concept thatprovides the possibility to connect resources to larger sys-tems. The distributed system consists of device services thatprovide features and functionalities to systems, and controlservices that create smart functionalities for the whole sys-tem. Different kinds of devices can be tested on the systemas they communicate similarly, and even complex applica-tions can be built rather easily using higher level controlservices. Dynamic capabilities of the general interface pro-vide the possibility to expand and adapt the system to matchthe task and environment’s requirements.

The other main objective here is to develop an architecturethat provides a software library that integrates all our re-search work to a rich set of tools. Therefore these methodsand solutions are easily usable on forthcoming projectswhich speeds up and develops our research work and capa-bilities.

Developing hardware architecture for a smart environment.

The aim is to develop a basis for devices operating in asmart environment. We have recently developed a modularelectronic concept, called Atomi, and created general soft-ware components for building complex activities. Atomielectronics provides us with a possibility to modify the hard-ware easily; new functionalities are available immediatelyin a plug-n-play way. Modularity has been achieved usingAtomi electronics and a modular software architecture -Property Service, which provides an interface for these re-sources and an opportunity to control different robots and

other devices in a smart environment.

Building a testing environment. In addition to software andhardware architectures, a test environment will be needed.We are building a prototype of a smart living room in ourlaboratory. This living room will be a large, proactive sys-tem. The technology and computation will be hidden as faras possible; the goal is to provide the user with a naturalenvironment that offers advanced services. The hardwarefor the first version of the prototype has been installed. Theimplementation of the basic software and the developmentof the computational methods will continue.

Remote Distributed Intelligence. The aim is to develop aprototype for testing and verifying Remote Distributed In-telligence. This is realized by implementing a multi-agentsystem composed of state-of-art miniature mobile robotsequipped with sensors and communication devices whichsupport a wide range of applications including guarding,inspection, and environment monitoring. This research willgive us valuable knowledge about the possibilities of a multi-robot team in real world applications.

The miniaturization of mobile robots is rapidly progressingdue to developments in electronics and material technol-ogy. Multirobot systems composed of a group of miniatur-ized robots can have numerous applications in the future.They may be a regular sight on space expeditions or theymay operate inside the human body for our well being. Atpresent, multirobot systems are a rare sight in real worldapplications. However, they have already been used in sur-veillance applications, military demonstrations, and in dis-tributed sensing applications.

Distributed sensing is an interesting application domain asthe multirobot system is inherently spatially distributed. Thespatial distribution of robots can be naturally utilized toovercome problems that the miniaturization of the robotsintroduces to the sensing technology in some problem do-mains. We have shown how a multirobot system can beused to extract 3-D information from an environment witha system of two heterogeneous robots which together forma distributed structured light based scanning device. Thedistributed 3-D scanning concept is useful in applicationswhere a group of robots needs to extract geometrical infor-mation in remote locations without a requirement that a sin-gle robot must alone carry the necessary technology forperforming the range measurements.

Miniature mobile robots form a multi-robot team to investi-

gate Remote Distributed Intelligence.


An infrared based location module was developed for theminiature robots. The module provides information aboutthe relative poses of the robots. The module can be used tocreate measurement formations and automatically performmeasurements about lighting and the magnetic field of theenvironment, for example.

Top: a range image from a cooperative 3-D scanning ex-

periment. Two robots have scanned a person laying on the

floor. The image demonstrates how, by simple coopera-

tion, a pair of miniature mobile robots (bottom) can obtain

dense range data from objects of an environment.

Wireless sensors. Based on our long experience in mobilerobotics and sensor systems, we have developed small wear-able wireless sensors for measuring the activity of humansand objects. The sensors provide information about 3-Dacceleration, temperature, atmospheric pressure, soundspace and lighting conditions every 10 ms. This informa-tion is used to recognize the context of a person. The sen-sors can also be attached to the objects the person ismanipulating, which makes it possible to register the meas-urements from different sources (human and objects), andto recognize various human-environment interactions. Rec-ognizing human-environment interactions has many appli-cations, especially in sports and education.

Swarm robotics. ROBOSWARM is an EU project that waslaunched on November 1st 2006. Together with eight otherparticipants ISG has an important role in this project, whichaims to develop an open knowledge environment for self-configurable, low-cost and robust robot swarms usable ineveryday applications. Advances in the state-of-the art ofnetworked robotics are proposed through the introductionof a local and global knowledge base for ad hoc communi-cation within a low-cost swarm of autonomous robots op-erating in a surrounding smart IT infrastructure.

Research on context-aware services

A growing variety of services are nowadays available forusers. We are approaching a situation where the sheernumber of services hinders their utilization. Our hypoth-esis is that this service overload can be managed by utiliz-ing the user’s context. To develop context-aware services,we have set and tackled the following objectives:

Recognizing the user’s context. The aim is to find and de-velop methods for recognizing the user’s context from sen-sor data using signal processing, pattern recognition andmachine learning methods. Context can be used to identifythe services that are relevant in the situation at hand - andto adapt these services. In earlier research, our focus hasbeen on wlan position data, data produced by our pressure-sensitive EMFI floor and heart rate measurements.

The research of user modelling on the EMFI floor has beenconcentrating on development of machine learning and pat-tern recognition methods for footstep-based person identi-

Miniature robot

with the location

module on the

top.

A wireless sensor (top right) and a base station (centre).

The base station receives the sensor readings and saves

them into a memory card. The data can later be downloaded

into a host computer for data analysis.

Measured di-

rections of the

local magnetic

field.

A measurement

formation.


fication as well for real-time multiple object tracking. Inthe person identification domain, different base classifiersfrom the statistical machine learning field are tested. Anensemble learning method, which uses multiple base clas-sifiers trained on different features sets, is developed. Inensemble learning, two level combinations of base classifi-ers’ probabilistic outputs and multiple adjacent footstep pro-files are fused to make an accurate decision on a person’sidentity. The best results were achieved using Support Vec-tor Machine (SVM) base classifiers and the features basedon spatial, statistical, and spectral properties of footsteppatterns. A 95% success rate of ten different walkers wasachieved. In addition, we have studied real-time multipleperson and mobile robot tracking, based on dynamic se-quential modelling of moving objects; they are individu-ally tracked using Particle Filtering (PF). The method isable to track a small number of objects simultaneously onthe floor and it gives ground for future work on higher levelcontextual and behavioural modelling of users and robots.The figure below shows an example of PF tracking esti-mates of two moving objects on the floor.

We have also continued our work on activity recognitionand sentient artefacts based on signal processing and learn-ing methods utilizing the measurements from different sen-sory devices. One of our post doc researchers visited theDistributed Computing Laboratory (DCL) at Waseda Uni-versity in Tokyo, Japan. Also, during 29th October - 28thDecember 2006, Dr. Kaori Fujinami from DCL visited ISG.

The research aims at reliable context recognition, utilizingdifferent sensory devices and sentient artefacts. Togetherwith DCL members, pattern recognition methods for ex-tracting people’s context, the management of contextual in-formation, and ways of dealing with the heterogeneity ofsensing device and contextual information are studied. Inour research, we have combined the environmental contextand the user context (activity). In our system, the user’sactivity was recognized from wearable 3D acceleration sen-sors (on the right wrist and thigh, triaxial acceleration sig-nals), shown here in thefigure on the right, with aneural network. This infor-mation was combined to thestate-of-use of a sentient ar-tefact (embedded with 3D ac-celeration sensors, pressuresensors, or other sensors)with a linkage condition (sit-ting activity linked to a chair,walking activity linked to adoor). This way the wearerwas identified as the user ofthe artefact. In the case ofmultiple users performing thesame activity, the identity ofthe user was verified with cor-relation analysis between theartefact’s and user’s accelera-tion signals.

Modelling user behaviour. The aim is to find models thatcan be used to predict user behaviour. User models willenable services to be adapted automatically based on theusers’ personal preferences and also enable proactiveness.Objectives will create the necessary basis for this work; themodel will be created by observing the user’s contextchanges and actions over time. In earlier research, we havecreated rule-based models of user behaviour by utilizingdata mining algorithms to associate different contexts inorder to better serve the user. In the future, adaptive ma-chine learning models based on neural networks, kernel

Particle Filtering tracking estimates of two moving objects

on the floor.

Knowledge environment for Interacting Swarms: demonstration system of ROBOSWARM.


VCG-based simulation of cardiac electrophysiologic func-

tioning.

methods, and sequential modelling will be applied alongwith the study of novel sensor measurements recorded fromthe different wearable and environmental sensing devices.We have continued this co-operation with Waseda Univer-sity, and also the co-operation with the Tokyo University ofAgriculture and Technology is planned.

Labelling personal information with context. The aim is todevelop methods that could help in the management andutilization of the growing quantities of digital informationrelated to our lives. The hypothesis is that context is anintuitive way of organizing personal information, as it al-lows users to relate information to their experiences. Weespecially aim to produce a methodology for collecting andstoring an extensive set of digitally available, context-la-belled data from a person’s everyday actions and activities,and to outline data mining and learning methods for ac-cessing, visualising and utilising such data in useful ways.

Providing easy access to services through tangible inter-

faces. The aim is to develop an intuitive tangible interfacefor requesting services with a single touch. We utilize RFIDtags mounted in the environment and RFID readers inte-grated in mobile phones. The tags connect the physical anddigital environments. Visual symbols communicate to us-ers the objects that can be touched and the services that canbe activated. When a user touches such a symbol with amobile phone, the data stored in the tag and other contex-tual information related to the situation trigger the requestedservice. This work is related to the first objective, recog-nizing the user’s context, as touching an RFID tag producesaccurate information about that context. We have designedsets of visual symbols in co-operation with the Departmentof Industrial Design at the University of Lapland. In theCAPNET project, we implemented together a tourist infor-mation board that is equipped with RFID tags and visualsymbols, see the figure below. Touching a symbol brings alist of services into the mobile phone’s screen. From thislist, the user can, for example, request a map to be shownon the phone’s screen. The map shows a path from the us-er’s current location to the target that she touched with thephone. The usability and user experience studies indicatethat touching visual symbols with a mobile device is aneasy way to activate services. Based on our findings, welist security, controllability and social acceptance as the mainchallenges in deploying this interaction method in every-day use.

Research on data mining

Biosignal processing

Sudden cardiac arrest is the most common cause of deathin western countries. It accounts for approximately 50% ofcardiovascular deaths and apparently has a highly variablepathophysiological etiology. The risk of sudden cardiac ar-rest is high in certain subgroups of patients with a historyof myocardial infarction and depressed left ventricular func-tion. The key question in research is why do some subjectsdevelop ventricular fibrillation during acute coronary oc-clusion, while others survive this episode without fatal ar-rhythmia. The challenge for research is to developapproaches or techniques that will allow the screening ofthe specific risk for fatal ventricular arrhythmias as a pre-dictor of the first event in patient populations that have alow cumulative risk, but generate a large number of vic-tims. In addition, the predictive value of many known riskfactors of sudden cardiac arrest among patients with knownheart disease has not been definitively established.

Our approach to studying the heart is based on the analysisof a 12-lead ECG (electrocardiogram). ECG contains in-formation on cardiac function that is highly relevant to di-agnosis and treatment. This information includes signals ofatrial electrical activity (P-wave of ECG and the correspond-ing atrial repolarization signal mixed with the QRS com-plex) and also of ventricular repolarization (T-wave). Wehave set and tackled the following objectives:

Computation of risk markers from ECG characteristics. Theaim is to develop new methods and algorithms for the analy-sis and interpretation of electrophysiological signals andthe autonomic regulation of the cardiovascular system. Thedigital 12-lead ECG was decomposed into multi-dimen-sional dipolar and non-dipolar components in the depolari-zation and repolarization phases in order to derive new riskmarkers of heart diseases. 12-dimensional principal com-ponent analysis was performed to extract dipolar (vectorcardiographic) and non-dipolar components from the meas-urement signal. The depolarization and repolarization phaseswere analysed separately, and various parameters (e.g., VCGloop descriptors and singular value-based descriptors) weredeveloped that describe the electrical functioning of the heartduring one beat or in continuous ECG.

Real-time measurement and interpretation of ECG includeschallenging signal processing problems that are related tofalse alarms caused by motion artefacts of the patient. New

Services can be activated by touching a tourist information

board with a mobile phone that is equipped with an RFID

reader.


adaptive signal processing solutions were developed thatfilter motion artefacts effectively through VCG-based mod-elling of heart behaviour, and produce significantly en-hanced ECG signals. Novel solutions for automaticallydetecting ECG morphology changes were derived from atheory of invariant pattern recognition. Techniques weredeveloped that can correct multichannel ECG for motionartefacts of moving patients. Algorithms were implementedthat can classify the motion-artefact corrected ECG in vari-ous classes depending on patient symptoms.

An ELD-based (Equivalent Double Layer) simulation en-vironment was utilized to model myocyte-level action po-tential changes of infarcted heart ventricle tissue. The infarctlocation, size and severity were varied and it was shownthat these parameters affect significantly some commonlyused ECG indexes.

Modelling of the functioning of the autonomic nervous sys-

tem. The relationships between ECG, blood pressure, breath-ing and sympathetic nervous activity signals were modelledand identified from actual signals in order to develop newmodels for the autonomic control of the cardiovascular sys-tem. A novel method was developed for adaptively filter-ing respiration artefacts from ECG and blood-pressuresignals. This method produces a significantly improved ofBSR estimate than earlier reported in the literature. Meas-urement equipment was used that only a few research groupspossess in the world.

Modelling of physical fitness. A new model was developedfor estimating the endurance fitness level of a runner fromthe physiologic signals and contextual information relatedto the runner. The new method can estimate the endurancefitness within 5% inaccuracy as compared to a running er-gometer.

Estimation methods of anaesthesia depth. The aim of thistask is to develop new methods and algorithms for estimat-ing the depth of anaesthesia from multi-channel EEG. Anew concept of relative induction time was developed whichenables a significant improvement in the depth estimationfrom EEG. The new method produces a time-continuousvalue for the depth estimate, and accurately predicts theinstance of loss of consciousness. The method was extendedsuch that it can model the depth of anaesthesia such depththat burst-suppression of EEG begins to evolve.

Modelling of human text writing. New techniques for pre-dicting text production of a keyboard user were developed.The solution includes advanced statistical models of wordcombinations and word-class combinations, which effec-tively enhance text production.

Real-time data mining systems

The goal of this research track is to create on-line data min-ing systems with semi-independent updating properties,methods for fast and reliable implementation of the sys-tems and installations of the new technology in real appli-cations.

Smart Archive. The aim is to develop a framework for real-time modelling of phenomena that continuously producenew measurement data. The focus area of the technologyunder development is intelligent on-line data processing forthe purpose of on-line modelling. The development of asoftware framework that enables the rapid realisation of newmodels as software applications is currently in progress.By using the framework, the implementation and set-uptimes of models can be significantly reduced, and their op-erational reliability improved. A specific issue to be ad-dressed is the selective storage and effective utilisation ofaccumulating data that eventually comes to represent thefull spectrum of the phenomenon being modelled. The tech-nologies to be developed are collectively known as the SmartArchive. An advanced prototype version is being used tocreate a series of data mining solutions for diverse indus-trial application domains, and based on the resulting expe-rience, the design of the framework will be further refined.The ultimate goal is a truly generic data mining meta-appli-cation with a graphical user interface that allows applica-tion instances to be constructed parametrically with theminimum quantity of application-specific code.

Joint modelling of mean and dispersion, based on large

data sets is an example of a real-time industrial applica-tion. Industrial applications of data mining often focus ondeveloping and optimizing production processes based ondata gathered from production lines. The benefits include a

Estimation of anaesthesia depth from ECG.

Analysis of an autonomic nervous system from multiple

biosignals.


higher level of automation, more accurate process control,better quality of goods, and improved customer satisfac-tion. The aim of this application was to develop new meth-odology for joint modelling of mean and dispersion forindustrial purposes. Models for predicting the mean andvariance of the mechanical properties (strength, elongation,toughness) of steel plate products have been developed. Themodels are utilized in a production line to optimize the proc-ess settings, based on the predicted risk of disqualificationin the testing of mechanical properties. Also, methodologyfor the modelling of a time-varying conditional variancefunction has been developed.

Research on software security

Within the Intelligent Systems Group, the Oulu UniversitySecure Programming Group(OUSPG) has continued re-search in the field of implemen-tation level security issues andsoftware security testing. Soft-ware implementation may in-troduce potential forunanticipated and undesiredprogram behaviour; for exam-ple, an intruder can exploit suchvulnerability to compromise thecomputer system.

In 2006, the research focus hasbeen on the implications and re-lationships of the different tech-nologies in complex systems.Information network environ-ments are more complex thanever before, and this complex-ity will increase in the future.One important factor affectingthe increase of the complexityis the convergence of informa-tion networks. One seemingly

simple event may generate a number of small actions indifferent parts of the network. Additionally, different com-ponents may use varying protocols. Thus the development,deployment and management of complex networks are la-borious and require in-depth understanding of differentfields. OUSPG has approached the issue from three differ-ent directions: causal relationships, protocol genes and pro-tocol dependence.

Causal relationships. The aim is to develop methods forinferring causal relationships in complex systems. The re-search applied data mining to network traffic to find datarelevant to the system being analysed. Applications of themethod include getting an overall view of the communica-tion patterns of complex systems, diagnostics, and securityrisk assessment.

Visualisations of file format analysis and fuzzing. Impact of several vulnerabilities.

Test cases test12.arj and test13.arj causing an implementation to fail.


The results of the project were commercialized in a spin-off company, Clarified Networks Oy. The business plan wasa finalist in the Venture Cup 2006 competition and wasawarded a local prize. The research results were also ap-plied in the HowNetWorks network monitoring appliance,which participated in the VMware Ultimate Virtual Appli-ance Challenge competition, and ultimately won the grandprize of $100,000.

In 2007, the work will be continued with a three-monthresearch visit to Dartmouth College, where the methods willdeveloped further for analysis of large scale wireless net-works.

Identification of protocol genes. This research approachesthe problem of complexity from the other direction by de-veloping tools and techniques for reverse-engineering andidentification of protocols using protocol genes - the basicbuilding blocks of protocols. The approach is to use tech-niques developed for bioinformatics and artificial intelli-gence. Samples of protocols and file formats are used toinfer structure from the data. This structural informationcan then be used to effectively create large numbers of testcases for this protocol.

Vulnerability management of the information infrastruc-

ture contributes to protocol dependence: another angle ofapproaching this problem is through technology dependen-cies. This activity studies the impact factor of different tech-nologies on CNI and develops a visual model forunderstanding dependencies related to protocols. This wasaccomplished by extending traditional Wikis, which are ef-fective mass collaborative authoring services, with graphingextensions. Graphingwiki enables the deepened analysis ofthe Wiki data by augmenting it with semantic data in a sim-ple, practical and easy-to-use manner. Visualisation toolsare used to clarify the resulting body of knowledge so thatonly the data essential for a usage scenario is displayed.Logic inference rules can be applied to the data to performautomated reasoning based on the data. Perceiving depend-encies among network protocols presents an example usecase of the framework.

Exploitation of Results

The results of our research were applied to real-world prob-lems in many projects, often in collaboration with indus-trial and other partners. Some examples of exploitation aredescribed below.

The Intelligent Systems Group utilizes a robotics labora-tory and pressure-sensitive floor (EMFi material) installedin our laboratory as part of a smart living room. Other equip-ment includes a home theatre, two degree-of-freedom ac-tive cameras, four mobile robots and one manipulator, aWLAN network, and various mobile devices (PDAs, a tab-let PC, Symbian mobile phones). WLAN positioning cov-ers a large part of the campus (including the laboratory),and a home automation network is being installed. Our aimis to gradually build a versatile infrastructure that offersvarious generic services for pervasive applications. Natu-rally, this kind of environment enables realistic experimentsthat lead to a better understanding of such applications.

The group’s expertise in robotics was applied in develop-ing a mobile robot for domestic help. A teleoperated robotserves as the remote eyes of the elderly and those who takecare of them. During the reporting period, the main taskwas to develop teleoperation capabilities for the robot. Avoice controlled service robot was successfully demon-strated. The purpose of the robot is to assist elderly peoplein their homes and provide a communication link to healthcare personnel. A design project was launched with theUniversity of Lapland to further develop the appearance ofthe robot, and make it suitable for various applications andresearch studies regarding human-robot interaction.

The development in robotics has continued in the area ofmechanical and miniaturization research. Qutie is an inter-active mobile robot designed in co-operation with the Uni-versity of Lapland. In the current year, development hasfocused on modular electronics, the Atomi concept, and thecreation of general software components for building com-plex activities for the robot. As the robot has several actua-tors and sensors it is also a good platform for developing amethod of building complex robots using modules.Modularity has been achieved using Atomi electronics andthe modular software architecture Property Service.

The Atomi electronics is an implementation of our Embed-ded Object Concept (EOC). The EOC utilizes commonobject-oriented methods used in software by applying themin combined Lego-like software-hardware entities, embed-ded objects. This concept enables fast prototyping with tar-get hardware, incremental device development andhigh-level device building for non-experts. The figure be-low demonstrates the idea of using the Atomi objects.

The development of a miniature robot is preparatory actiontowards swarm robotics research. Instead of using large

Prototyping with Atomi objects.


robots, it is often desirable to have multiple small robots tosave valuable work space and make the maintenance of therobots easier. Also, the implementation cost of a miniaturerobot is lower because of the simpler mechanical design.We have developed a novel modular miniature mobile ro-bot designed for swarm robotics research. The sensor set ofthe robot includes a colour stereo camera system with twoCMOS cameras and DSP, allowing each robot to performsophisticated stereo image processing on-board. The modu-lar design permits the addition of new modules into the sys-tem. The modules communicate using three serial buses(SPI, I2C, and UART), which enable flexible, adaptive, andfast inter-module data exchange. The robot is developedfor swarm robotics research with the aim of providing alow-cost and low-power miniature mobile robot with capa-bilities typically found only in large robots.

During year 2006 a group of five miniature robots werefabricated, and all the necessary software components wereimplemented to enable the utilization of the multi-robot teamin basic research.

The robot developed is a part of research which aims atunderstanding how a global objective can be achieved by amulti-agent system without explicit regard for cooperationwith the other agents, and to investigate the relationship ofspatial patterns composed of interacting entities (agents)and the resulting dynamics.

In addition, we are investigating how humans can effort-lessly interact and control a multi-agent system and gainmeaningful information about the environment through it,and we study what are the minimal requirements for an agentin order to produce useful behaviour at the system level.

In addition to the core research, OUSPG also maintains atest network infrastructure for research groups such asOUSPG, ISG and MediaTeam. Test networks are requiredfor the safety of both our own research and innocent by-standers, and one was constructed for internal use in thelaboratory. The network provides a fully functional infra-structure with services such as storage, backups, DNS andmail.

The infrastructure started life as an isolated test networkfor OUSPG use, but has expanded to provide a basic infra-structure for others with similar needs. There are three sepa-rate networks: a core network for basic services, thecombined playground and the wireless access networkpanOULU, and an Internet-connected network with con-nectivity independent of the main university network.

Future Goals

We will continue to strengthen our long term research andresearcher training. We will also continuously seek oppor-tunities for the exploitation of our research results by col-laborating with partners from industry and other researchinstitutions in national and international research programsand projects. The group is a founding member of the Euro-pean Robotic Network of Excellence (EURON). The groupis a contract member of EURON II which was approvedfor the EU’s 6th framework as a Network of Excellence. Wewill strengthen our international research co-operation. One

of our researchers made a 10 month post doc research visitto Professor Tatsuo Nakajima at Waseda University. Weare making arrangement for another researcher visit to hislaboratory. Also a three-month research visit to DartmouthCollege will be made in the Spring 2007 and one of ourresearchers will start a one-year visit to NRC Cambridge inAutumn 2007. The group is participating two EC fundedprojects. A STREP project, ROBOSWARM had its kick-off meeting November 2006 and an IP project XPRESSwill start in January 2007.

Personnel

srotcod&srosseforp 9

stnedutsetaudarg 32

srehto 33

latot 56

sraeynosrep 35

External Funding

ecruoS RUE

dnalniFfoymedacA 00083

noitacudEfoyrtsiniM 000232

sekeT 000764

etavirpcitsemod 000752

lanoitanretnirehto+UE 000623

latot 0008331

Doctoral Theses

Laurinen P (2006) A top-down approach for creating and imple-menting data mining solutions, Acta Universitatis Ouluensis C246.

Koskinen M (ISG) Automatic assessment of functional suppres-sion of the central nervous system due to propofol anesthetic in-fusion. From EEG phenomena to a quantitative index. ActaUniversitatis Ouluensis C 253.

Juutilainen I (2006) Modelling of conditional variance and un-certainty using industrial process data. Acta Universitatis OuluensisC 258.

Kerttula M (2006) A framework for the development of personalelectronic products. University of Oulu.

Selected Publications

Ahonen P, Karjalainen K, Kuusela E, Lehtonen S, Puuperä R,Röning J, Savola R, Tokola T & Uusitalo I (2006) Informationsecurity in wireless networks. LUOTI-julkaisuja 9/2006.

Elsilä U & Röning J (2006) Analysis of wedge formation in hotstrip rolling after continuous casting. - Modeling of casting, weld-ing, and advanced solidification processes. Eds. C-A Candin &


M Bellet, USA. The minerals, metals & materials society, 767-774.

Eronen J & Röning J (2006) Graphingwiki - a Semantic Wikiextension for visualising and inferring protocol dependency. Proc.1st SemWiki Workshop, June 11-14, Budva, Montenegro.

Fujinami K, Pirttikangas S, Nakajima T (2006) Who opened thedoor?: Towards the implicit user identification for sentient arte-facts Adjunct Proceedings of Pervasive 2006, May 7-10, Dublin,Ireland, 107-111.

Haapalainen E, Laurinen P, Junno H, Tuovinen L & Röning J(2006) Feature Selection for identification of spot welding proc-esses. The 3rd International Conference on Informatics in Con-trol, Automation and Robotics (ICINCO 2006), August 1-5,Setubal, Portugal.

Hautala A, Kiviniemi A, Mäkikallio T, Tiinanen S, Seppänen T,Huikuri H & Tulppo M (2006) Peripheral sympathetic outflowcorrelates with the response to endurance training. 11th Euro-pean Collage of Sport Science (ECSS), Lausanne, Switzerland,Book of Abstracts p. 304.

Juutilainen I, Röning J (2006) Adaptive modelling of conditionalvariance function. 17th Symposium of IASC (COMPSTAT 2006),August 28 - September 1, Rome, Italy.

Juutilainen I & Röning J (2006) Planning of strength marginsusing joint modelling of mean and dispersion. Materials and Manu-facturing Processes 21(4): 367-373.

Kajava J, Anttila J, Varonen, R, Savola R & Röning J (2006) In-formation security standards and global business. Proceedings ofInternational Conference on Industrial Technology (ICIT 2006),December 15-17, Mumbai, India, 2091-2095.

Kajava J, Anttila J, Varonen R, Savola R & Röning J (2006) Sen-ior executives commitment to information security - from moti-vation to responsibility. International Conference onComputational Intelligence and Security (CIS2006), November3-6, Guangzhou, China, 1519-1522.

Kawsar F, Fujinami K, Pirttikangas S, Nakajima T (2006) Per-sonalization and context aware services: A middleware perspec-tive. Proc. 2nd International Workshop on Personalized ContextModeling and Management for UbiComp Applications, in con-junction with UbiComp2005, the 7th International Conferenceon Ubiquitous Computing.

Kemppainen A, Haverinen J & Röning J (2006) An infrared loca-tion system for relative pose estimation of robots. Proc. 16-thCISM-IFToMM Syposium of Robot Design, Dynamics, and Con-trol (ROMANSY 2006), June 20-24, Warsaw, Poland, 379-386.

Kiviniemi A, Hautala A, Mäkikallio T, Seppänen T, Huikuri H &Tulppo M (2006) Cardiac vagal outflow after aerobic training byanalysis of high-frequency oscillation of the R-R interval. Euro-pean Journal of Applied Physiology, 96(6):686-692.

Kiviniemi A, Huikuri H, Hautala A, Tiinanen S, Seppänen T,Mäkikallio T, Tulppo M (2006) Changes in fluctuation pattern ofcardiovascular signals during sympathetic activation. 11th Euro-pean Collage of Sport Science (ECSS) Annual Meeting, Lausanne,Switzerland 2006. Book of Abstracts, p. 301.

Koskinen M, Mustola S & Seppänen T (2006) Forecasting theunresponsiveness to verbal command on the basis of EEG fre-quency progression during anesthetic induction with propofol.IEEE Transactions on Biomedical Engineering 53(10): 2008-2014.

Koskinen M, Seppänen T, Tong S, Mustola S & Thakor N (2006)Monotonicity of approximate entropy during transition fromawareness to unresponsiveness due to propofol anesthetic induc-tion. IEEE Transactions on Biomedical Engineering 53(4): 669-675.

Laurinen P, Siirtola P & Röning J (2006) Efficient algorithm forcalculating similarity between trajectories containing an increas-

ing dimension. Artificial Intelligence and Applications (AIA 2006),February 13-16, Innsbruck, Austria.

Linna E, Perkiömäki J, Karsikas M, Seppänen T, Savolainen M,Kesäniemi A & Huikuri H (2006) Functional significance ofKCNH2 (HERG) K897T polymorphism on cardiac repolarizationassessed by analysis of T-wave morphology. Annals of NoninvasiveElectrocardiology 11: 57-62.

Perkiömäki J, Hyytinen-Oinas M, Karsikas M, Seppänen T,Hnatkova K, Malik M & Huikuri H (2006) Usefulness of T-waveloop and QRS complex loop to predict mortality after acute myo-cardial infarction. The American Journal of Cardiology. 97: 353-360.

Pietikäinen P & Huttunen L (2006) Behavioral study of bot obe-dience using causal relationship analysis. Proc. 18th Annual FIRSTConference, June 25-30, Baltimore, MD, USA.

Pirttikangas S, Fujinami K, Nakajima T (2006) Feature selectionand activity recognition from wearable sensors. International Sym-posium on Ubiquitous Computing Systems (UCS2006), October11-13, Seoul, Korea, 516-527.

Pirttikangas S, Riekki J & Hietajärvi J (2006) Extensions forCommonGIS to aid the development work of context-aware sys-tems. Advances in Computer Science and Technology (ACST2006), January 23-25, Puerto Vallarta, Mexico.

Riekki J, Salminen T & Alakärppä I (2006) Requesting pervasiveservices by touching RFID tags. IEEE Pervasive Computing5(1):40-46.

Seydou F, Ramahi O, Duraiswami R & Seppänen T (2006) A nu-merical computation of Green’s function in two-dimensional fi-nite-size photonic crystal of infinite length. Optics Express 14(23):11362-11371.

Seydou F, Ramahi O & Seppänen T (2006) Analytical computa-tion of energy levels and wave functions in chaotic cavities. In-vited speech. Bellagio International Workshop on MathematicalModeling, Simulation, Visualization and e-Learning, November20-26, Bellagio, Italy.

Seydou F & Seppänen T (2006) A volume integral method in in-verse scattering. IEEE AP-S International Symposium, USNC/URSI National Radio Science Meeting AMEREM Meeting, July9-14, 2006, Albuquerque, New Mexico.

Tikanmäki A, Haverinen J, Kemppainen A & Röning J (2006)Remote-operated robot swarm for measuring an environment.Proc. ICMA 2006 - International Conference on Machine Auto-mation, June 7-8, Seinäjoki, Finland.

Tulppo M, Kiviniemi A, Hautala A, Mäkikallio T, Tiinanen S,Seppänen T & Huikuri H (2006) Association between baroreflexsensitivity and aerobic fitness after dynamic exercise. ACSM An-nual Meeting, Med Sci Sports Exer. Suppl. 38: 18.

Vallius T & Röning J (2006) A telepresence robot system realizedby embedded object Concept. Proc. Intelligent Robots and Com-puter Vision XXIV: Algorithms, Techniques, and Active Vision,October 3-4, Boston, Massachusetts, USA.

Vallius T & Röning J (2006) ATOMI II - Framework for easybuilding of object-oriented embedded systems. Proc. 9thEuromicro Conference on Digital System Design: Architectures,Methods and Tools, August 30 - September 1, Dubrovnik, Croatia,464-472.

Wieser C & Röning J (2006) Über die Verwundbarkeit von IP-Telefonie-Systemen. Proc. 13th Workshop Sicherheit in vernetztenSystemen, March 1-2, Hamburg, Germany 2006.

Wieser C, Röning J & Takanen A (2006) Security analysis andexperiments for VoIP RTP media streams. Proc. 8th InternationalSymposium on Systems and Information Security (SSI2006).November 8-10, Sao Jose dos Campos, Sao Paolo, Brazil.

intelligent systems group (isg) - infotech oulu · 28 infotech oulu annual report 2006 intelligent...

Documents