Intelligent Technologies and Systems Research Team

Transcription

1 Intelligent Technologies and Systems Research Team Centre for Intelligent Technologies KKUI FEI TU v Košiciach B. Němcovej Košice Slovakia Team Members Prof. Peter Sinčák, PhD. Assoc. Prof. Marián Mach, PhD. Assoc. Prof. Kristína Machová, PhD. Assist. Prof. Ján Vaščák, PhD. PhD students: Tomáš Cádrik Gergely Magyar Martina Szabóová Jaroslav Ondo Peter Takáč Jakub Hvizdoš Martin Čertický Michal Puheim Research Direction Research activities are primary focused on three areas and their mutual intersection: Intelligent robotics the centre focuses on selected parts of symbolic as well as subsymbolic artificial intelligence (e.g., machine learning, neural networks, fuzzy systems, and evolutionary computation) in order to employ them in the field of robot systems for recognition, information processing and knowledge creation as well as

2 decision making and communication. Our long-term aim is to develop learnable collaborative robot systems. Intelligent space the aim is to develop a space equipped with suitable sensors, devices, information and communication technologies in such way that it can play the role of an interactive intelligent environment able to perceive and recognise activities and events performed within this space as well as to react on them, make decisions and influence the interior space. Cloud computation our ambition is to use scalable cloud technologies in the field of intelligent (e.g., robotic) systems. On one hand, it can be on the infrastructure level for communication, virtualisation and information as well as knowledge sharing, which enables to manage big groups of different devices in an easy way. On the other hand, it is possible on the level of so called AI bricks modular services providing functionality of selected artificial intelligence methods enabling an effective way how to add intelligent capabilities to technological devices. The Importance and Benefits of Research Research directions of the team follow modern trends in the field of ICT as well as areas of industry specialisation RIS3 SK, mainly automation, robotics, and digital technologies. Our research is fully compatible with the INDUSTRY 4.0 concept and its parts. The research team uses several robot platforms for its research activities the centre is equipped with 18 robots NAO, two mobile robots Q.bo and two humanoid robots Zeno by Hanson Robotics. Team members together with students (mainly PhD study level) employ for software application development newest technologies provided by Microsoft, which is one of leaders in the field of cloud services in the world. Deployment of cloud computation techniques in robotics creates preconditions for constructing light, cheap and intelligent robots, whose complex control and decision making mechanisms (their brain ) will be located and shared in cloud. These mechanisms represent knowledge base, behaviour planning, information processing, communication support and many others, which are essential for effective participation in solving real situations. The transition from individual devices (e.g., robots) to a space, in which these devices operate, represents the next evolutionary step. Exploitation of intelligent space technologies enables to solve more effectively such complex problems, in which devices of different types are forced to communicate with one another in order to solve tasks. The advantage is that constraints caused by limited views of particular devices are removed. Expertise transfer from the field of artificial intelligence into the field of the development of flexible robot systems and management of device spaces provides clear potential to develop the given fields also in those directions, in which currently dominates presence of human factors.

3 Solving Current Problems The research team participated in several problem solutions. A few samples: A teleoperation platform based on the Azure cloud platform has been developed in the collaboration with Italian firm Technodeal s.r.l. It uses the WEBAPI technology based on the HTTP protocol for communication. Teleoperation of several types of robots is supported the only compatibility condition requires running the Linux operation system on teleoperated robots. User can register robots and store different robot movements into a storage in order to use the movements during teleoperation. At the same time, user can search for and obtain movements suitable for a particular robot. The platform has been tested using several robots in several parts of the world: robots Zeno, Q.bo and NAO (teleoperation among Slovakia, Italy and Japan), robot Doro (teleoperation from Slovakia to Italy) and robot Pepper (teleoperation from Slovakia to Japan). Robots need to learn how to move. Thus, we have created a cloud-based system based on the conversion of human movement recordings acquired by Kinect sensor to movements for a humanoid robot. This research involved creating a local client application, which was designed to acquire data from the Kinect sensor and forward the data to the cloud environment. The data were in this environment interpreted and converted to data suitable for running on the NAO robot platform.

4 In order to solve problems in the domain of intelligent spaces, we have created a space enabling to determine localization and track position of persons and robots in this space based on the employment of visual and spatial information. The aim is to solve navigation tasks under dynamic changes of obstacles. The space with the area exceeding 200 m 2 is equipped with a set of 16 IP cameras connected to the local network and 6 Kinect sensors connected to minicomputers (allocated according to the following figure). In future, this intelligent space will be expanded with additional sensors of various types. It enables to scan all activities typical for indoor environments, which can be solved using automata or robots, including human-machine (robot) interaction as well as interaction among machines (robots), which can lead to cooperative robotics within the Internet of Things environment. Projects International projects: : Microsoft - Azure Machine Learning Award : Microsoft - Microsoft Azure Offer for Research Institutions : Virtlab, HUSKA Intereg project cooperation with BAYLOGi Hungary

5 National projects funded by Slovak funding schemes or agencies: : VEGA 1/0773/16 Artificial intelligence methods for intelligent robotics based on cloud computation. 2016: FEI grant System for robot teleoperation in intelligent space. 2016: Tatrabanka Coexistence of robots and humans in Internet of Things learning : KEGA 034TUKE-4/2014 Integration of study programmes Cybernetics and Artificial intelligence : VEGA 1/0667/12 Incremental learning methods for intelligent systems. 2014: Foundation Tatrabanka G.E.C.K. Project : VEGA 1/0042/10 Methods for identification, annotation, search, accessing and composition of services using semantic metadata to support selected process types : VEGA 1/0885/08 Learning systems based on computational intelligence. National projects funded by EU structural funds: : Technicom University Science Park TECHNICOM for innovative applications supported by knowledge technologies (ITMS ) : AS FEU, Package of quality improvement at TUKE through networks, ITMS: : AS FEU, Package add-ons for further reform of education at TUKE, ITMS: : AS FEU, Package of elements for improvement and innovation in education at TUKE, ITMS: : CE-FEI-II Development of the Centre for communication and information technologies for knowledge systems (ITMS ) : CE-FEI-I - The Centre for communication and information technologies for knowledge systems (ITMS ). Cooperation with Academic Institutions and Industry Scuola Superiore Sant Anna, Pisa, Italy, Prof. Cavallo Chukyo University Nagoya, Japan, Prof. Hartono Kyutech, Department of Human Intelligence, Kittakyushu, Japan, Prof. Wagatsuma, Prof. Yoshida (Memorandum of Understanding )

6 Waseda University, Humanoid Institute, Tokyo, Japan, Dr. Salvadore Sessa, Prof. Hashimoto (Memorandum of Understanding, ) ATR Japan, Advanced Telecommunication Institute, Japan, Prof. Kanda, Prof. Nishio University of Technology, Peking, China, Prof. Hirota KAIST, Korean Advanced Institute of Technology, Centre for Robot Intelligence, Prof. Joahim Kim University of Miskolc, Department of Informatics, Hungary, Prof. L. Kovacs, Prof. S. Kovacs BayLogi Institute, Department of Informatics, Miskolc, Hungary, Prof. Vas Loughborough University, Department of IT, UK, Prof. M. Zecca Faculty of electrical engineering and informatics, VŠB technical university Ostrava, Czech republic, Prof. Zelinka, Doc. Šaloun Faculty of informatics and management, University Hradec Králové, Czech republic, Dr. Mls University Pablo de Olavide, Sevilla, Spain, Prof. Salmeron Microsoft Slovakia, Bratislava, Dr. Roman Russev CEITGROUP a.s. Žilina, Slovakia, prof. Gregor Neulogy a.s., Bratislava, Slovakia, Dr. Lipnická ZTS VVU a.s. Košice, Slovakia, Dr. Jaroslav Jezdny Betamont s.r.o., Zvolen, Slovakia, Ing. Rastislav Babica Selected Publications Emergent Trends in Robotics and Intelligent Systems Where is Role of Intelligent Technologies in the Next Generation of Robots? / Peter Sinčák, Pitoyo Hartono, Maria Virčíkova, Ján Vaščák, Rudolf Jakša (Eds.), Spriger Verlag. Advances in Intelligent Systems and Computing, Volume 316, ISSN , 2015 Vehicle Navigation by Fuzzy Cognitive Maps Using Sonar and RFID Technologies / Ján Vaščák, Jakub Hvizdoš - In: SAMI Danvers : IEEE, 2016 P ISBN Theoretical Analysis of Recent Changes and Expectations in Intelligent Robotics / Peter Sinčák et al. - In: Advances in Intelligent Systems and Computing. Vol. 316 (2015), p ISBN ISSN Cloud-Based Image Recognition for Robots / Lorenčík Daniel et al.- In: Advances in Intelligence Systems and Computing. - Heidelberg : Springer, 2015 Vol. 345 (2015). - ISBN ISSN Computers Capable of Distinguishing Emotions in Text / Martina Tarhaničová, Kristína Machová, Peter Sinčák - In: Advances in Intelligent Systems and Computing.

7 - Switzerland : Springer, 2015 Vol. 316 (2015), p ISBN ISSN The method based on interference of waves for solving real value optimisation problems / Tomáš Cádrik, Marián Mach - In: Mendel Brno : University of Technology, 2015 P ISSN Computers Capable of Distinguishing Emotions in Text / Martina Tarhaničová, Kristína Machová, Peter Sinčák - In: Advances in Intelligent Systems and Computing. - Switzerland : Springer, 2015 Vol. 316 (2015), p ISBN ISSN Learning of Fuzzy Cognitive Maps by a PSO Algorithm for Movement Adjustment of Robots / Ján Vaščák, Roman Michna - In: Advances in Intelligent Systems and Computing. - Switzerland : Springer, 2014 Vol. 316 (2015), p ISSN Comparison Study of Robotic Middleware for Robotic Applications / Gergely Magyar, Peter Sinčák, Zoltán Krizsán - In: Advances in Intelligent Systems and Computing. Vol. 316 (2015), p ISSN Interference of Waves Based Usage of an Optimization Algorithm for Finding Rules in an Agent System / Tomáš Cádrik, Marián Mach, Peter Sinčák - In: SCIS and ISIS Danvers : IEEE, 2014 P ISBN Exracting rules from ZCS evolutionary classifier system for the purpose of future usage in robotic systems / Tomáš Cádrik, Marián Mach - In: Mendel Brno : University of Technology, 2014 P ISBN Cloud-Based Object Recognition: A System Proposal / Daniel Lorenčík, Martina Tarhaničová, Peter Sinčák - In: Robot Intelligence Technology and Applications 2 : Results from the 2nd Int.. Conference on Robot Intelligence Technology and Applications : Springer International Publishing, 2014 Vol. 274 (2014), p ISBN ISSN Personalized Emotional Expressions to Improve Natural Human-Humanoid Interaction / Mária Virčíková, Peter Sinčák, Dong Hwa Kim - In: Advances in Intelligent Systems and Computing. Vol. 208 (2013), p ISSN Real-time fuzzy logic-based hybrid robot path-planning strategies for a dynamic environment / Napoleon H. Reyes et al. - In: Efficiency and Scalability Methods for Computational Intellect. - Hershey : IGI Global, 2013 P ISBN Distributed and Incremental Visual Object Categorization for Humanoid Platform NAO / Peter Sinčák et al. - In: Advances in Intelligent Systems and Computing. Vol. 208 (2013), p ISBN ISSN

8 Adaptation of Fuzzy Cognitive Maps for Navigation Purposes by Migration Algorithms / Ján Vaščák, Martin Paľa - In: International Journal of Artificial Intelligence. Vol. 8, no. S12 (2012), p ISSN Towards the assisted teleoperation systems / Martin Paľa, Peter Sinčák - In: Ultra modern Telecommunications and Control Systems and Workshops (ICUMT) 2012 : 4. international congress : October 2012, St. Petersburg, Russia. - St. Petersburg : IEEE, 2012 P ISBN Adaptive Fuzzy Cognitive Maps Using Interactive Evolution: A Robust Solution for Navigation of Robots / Daniel Lorenčík, Ján Vaščák, Mária Virčíková - In: RiTA 2012 : Conference Proceedings : The 1st International Conference on Robot Intelligence Technology and Applications 2012 : Gwangju, Korea, December 16-18, Seoul : C-Agency, 2012 P Decision-making Systems in Mobile Robotics / Ján Vaščák - In: Autonomous Decision Systems Handbook. Praha, 2011 P ISBN Photos Laboratory of humanoid robots

9 Robot platforms Q.bo by TheCorpora (left) and NAO by Aldebaran (in the middle and right). Robot platform Zeno by Hanson Robotics