Projects

Current projects

mathisis_logoEU Horizon2020-ICT-20-2015: MaTHiSiS (Managing Affective-learning THrough Intelligent atoms and Smart InteractionS). Personalized, affect-enhanced smart learning, using robots, mobile devices and interactive boards (2016-2018).

http://mathisis-project.eu/

The MaTHiSiS learning vision is to provide a product-system for vocational training and mainstream education for both individuals with an intellectual disability and non-diagnosed ones. This product-system consists of an integrated anthropocentric platform, along with a set of learning components (educational material, digital educational artefacts etc.), which will respond to the needs of a future educational framework and provide capabilities for: i) adaptive learning, ii) automatic feedback, iii) automatic assessment of learner’s progress and behavioural state, iv) game-based learning. MaTHiSiS will make use of cutting-edge technologies in learning settings that range from specialized robots and mobile devices to interactive whiteboards, and will advance them to a greater degree of integration into the market. MaTHiSiS will create a novel and continuously adaptable “robot/machine/computer”-human interaction educational scheme which adapts to different learning requirements and makes use of the shared knowledge among its different components to advance learning beyond linear social skill acquisition towards more workplace-oriented activities.

Consortium:

  • ATOS SPAIN SA (ATOS) – ES
  • DIGINEXT SARL (DXT) – FR
  • Centre for Research and Technology Hellas (CERTH) – GR
  • National Center for Scientific Research”DEMOKRITOS” (NCSR) – GR
  • The Nottingham Trent University (NTU) – UK
  • University of East London (UEL) – UK
  • Universiteit Maastricht (UM) – NL
  • The University of Nottingham (UoN) – UK
  • Vrije Universiteit Brussel (VUB) – BE
  • Fondazione Mondo Digitale (FMD) – IT
  • Hellenic Telecommunication Company S.A.
  • Viesoji Istaiga Svietimo Ir Kulturos Mobiliuju Technologiju Institutas (IMOTEC) – LT
  • La Cometa Del Sud (LCS) – IT
  • NUROGAMES GMBH (NG) – DE
  • Consejeria de Educacion de la Junta de Castilla y Leon (JCYL) – ES
  • National Organisation for the Certification of Qualifications and Vocational Guidance (NOCQVG) – GR
  • Istituto Comprensivo Statale B. Lorenzi Fumane VR (PE) – IT
  • AEROSPACE VALLEY (AV) – FR

contact: stelios.asteriadis@maastrichtuniversity.nl

logoICTpequeEU Horizon2020-PHC-25-2015: ICT4Life (ICT services for Life Improvement For the Elderly). Ambient-Assisted Living for the elderly with cognitive impairments, personalized interfaces and teleservices (2016-2018).

http://ict4life.eu/

ICT4Life will develop a solution for individuals with early stage cognitive impairment living alone that will permit doctors and caregivers to extract information about patients (for taking the best medical or social actions), while contributing in a user friendly way to extend their independence. People with dementia, in general and, in particular, with Alzheimer at an early stage, and with Parkinson, constitute the main group of users ICT4Life will focus its analysis on. The technologies that will constitute the main pillars of ICT4Life are the following: Education delivery methods for care professionals; tools for the automatic creation of reusable resources, i.e. formative training content; ICT decision support instruments, incorporating machine intelligence, allowing care professionals to have access to treatment/therapy recommendations according to patients’ profile; advanced visual analytics to monitor personal health status; indoors activity recognition through advanced sensing to detect deviations from persons’ daily conduct; game-based learning and adaptation tools to boost patient empowerment and self-care abilities; integration of unobtrusive biomedical sensors for continuous monitoring of the patient health status and compilation data to be processed and generate new medical knowledge; design and development of personalised interfaces relying on smart/Connected TV and mobile technologies; standard codification mechanisms to increase levels of integration of clinical and social data from multiple service providers and user-generated data. Electronic Health/Care Record (EHR); data mining to accomplish data analytics so that new knowledge about co-morbidities, disease evolution and socio demographic parameters will be generated

Consortium:

  • Artica Telemedicina (ES)
  • Universidad Politecnica de Madrid (ES)
  • Asociacion Parkinson Madrid (ES)
  • Netis Informatics Ltd. (HU)
  • E-seniors: Initiation des Seniors aux NTIC (FR)
  • Centre for Research and Technology Hellas (EL)
  • Maastricht University (NL)
  • European Hospital and Healthcare Federation (BE)
  • University of Pecs – Medical School (HU)

 contact: stelios.asteriadis@maastrichtuniversity.nl

brain_robot_logo 2015 – 2018: “Advanced Brain-Robot Interfaces”, funded with UM’s Luik3 program. This is a joint project with Prof. Rainer Goebel and Dr. Bettina Sorger from the Department of Neuroscience of Maastricht University.

Project website: project.dke.maastrichtuniversity.nl/brain-robot
Contact: gerhard.weiss@maastrichtuniversity.nl

MISS_logo

2012 – 2016: RAI is leading the MiSS project which is a part of CATCH program.  MiSS aims at development of practical identity resolution techniques for Dutch historical archives, and extraction of social networks from large entity networks.

contact: gerhard.weiss@maastrichtuniversity.nl

ThinkSlim_logo_300pxphilips 2012 – 2016: RAI is participating in the Think Slim project which is a part of Healthy Life Style (HLS) partnership programme Think Slim aims to develop a tailored E‐coach­‐intervention for overweight people that is based on cognitive behavioral therapy (CBT). This is a joint project with the Faculty of Psychology and Neuroscience (FPN) of Maastricht University.

contact: gerhard.weiss@maastrichtuniversity.nl

Previous projects

  • Multi-Agent Learning in Auctions: The Design and Analysis of Markets and Traders
    • Funding: NWO Toptalent grant
    • PhD. student: M. Kaisers
    • Other researchers involved: K. Tuyls; S. Parsons; R. Peeters; D. Hennes
    • Period: September 2008 – September 2012
    • Short description: This project investigates how machine learning techniques can help to analyze and improve markets and trading strategies within the domain of auctions. An empirical approach allows new insights into auctions, which grow in importance in resource allocation problems and e-commerce. Key techniques investigated are replicator dynamics of evolutionary game theory and reinforcement learning.
  • Bee Swarming in Multi-Agent Systems
    • Funding: tUL grant
    • PhD. student: N. Lemmens
    • Other researchers involved: K. Tuyls, G. Weiss, A. Nowe, S. de Jong
    • Period: December 2006 – December 2010
    • Short description: This project investigates a new line of research within the field of swarm intelligence, i.e. honeybee inspired algorithms and their application to Multi-Agent Systems for solving optimization problems. Besides bee-inspired algorithms, hybrid algorithms (a combination of ant and bee behavior) will be developed and applied to simulated foraging and mobile ad hoc routing. Mobile Ad-hoc Networks pose a challenging problem to routing protocols. They consist of nodes with high mobility and there is no preset infrastructure available.
  • Hybrid Human-Agent Networks
    • Funding: Casimir – Dutch ministry of economic affairs
    • Phd. student: Xiaoyu Mao
    • Other researchers involved: N. Roos
    • Period: September 2005 – September 2010
    • Short description: The research in the Hybrid Human-Agent Networks project focuses on the application of Dynamic Distributed Constraint Satisfaction Problems for planning. The general idea is that a network of humans and agents collaboratively create a plan, dynamically refine the plan during execution, and handle incidences. The human actors in the network will choose between alternative options and will be able to introduce and withdraw constraints. The application domain of the project is the arrival, departure and taxi planning on an airport. The Dutch National Aerospace Laboratory (NLR) supports the project by providing the necessary domain knowledge.
  • Automatic Mental Health Assistant – AMHA
    • Funding: NWO/ESA (Joint project with Eindhoven University of Technology)
    • Researcher involved UM: K. Tuyls
    • Period: July 2008 – July 2012
    • Short description: This project investigates the psychological factors influencing astronauts on a long flight mission to Mars using techniques from (evolutionary) game theory, psychology and artificial intelligence. The AMHA project is part of the Mars-500 experiment which is carried out at the Institute for Biomedical Problems (IBMP) in Moscow. The European Space Agency (ESA) and the Russian Academy of Sciences jointly plan and conduct this experiment in order to simulate a manned mission to Mars. It will provide the unique opportunity to study interactions between crewmembers while collecting data about their health and performance during experimental isolation. The confinement study will imitate all key peculiarities expected to be present during future missions to Mars (i.e. ultra long duration flight, need for autonomy, complicated communication due to signal delay, and limited stock of expendables).
    • More info:http://www.amha.id.tue.nl/
  • Multi-Agent Relational Reinforcement Learning
    • Funding: ICIS grant of Dutch ministry of economic affairs(2006-2009)/ UM(2009-2010)
    • PhD. student: M. Ponsen
    • Other researchers involved: K. Tuyls, J. Ramon, M. Kaisers
    • Period: September 2006 – September 2010
    • Short description: Researchers in machine learning and adaptive systems have been addressing issues concerned with learning and adapting from past experience, observation, failures, opponent behavior etc. Whereas most of this research has focused on techniques for acquisition and effective use of problem solving knowledge from the viewpoint of a single autonomous agent, recent investigations have opened the possibility of application of some of these techniques in multi-agent settings. This project is the further development of a recent new line of research in this area, i.e. multi-agent relational reinforcement learning in a joint effort between experts from both multi-agent learning and relational reinforcement learning. Furthermore the connection to evolutionary game theory is investigated, as well as the application to the game of Poker.
  • Evolutionary Dynamics of Multi-Agent Learning
    • Funding: Maastricht University
    • PhD. student: D. Hennes
    • Other researchers involved: K. Tuyls, G. Weiss, M. Kaisers
    • Period: September 2009 – September 2012
    • Short description: Multi-Agent Systems (MAS) are accepted to be an important method for solving problems of a distributed nature. Key to the success of MAS is efficient and effective Multi-Agent Learning (MAL). In this project we address two important research challenges in MAL. 1) Current learning theory for single agents does not extend to MAL. Convergence guarantees no longer hold and there exists no general formal theory describing and elucidating the conditions under which algorithms for MAL are successful. 2) It is currently an open question how to scale up MAL, i.e. how to efficiently handle many states, many actions and many agents. We address both challenges using an innovative combination of Evolutionary Game Theory, Reinforcement Learning and Dynamical Systems Theory. More precisely, we provide a theoretical backbone for MAL based on the replicator equations of EGT.
  • Agent-based Interoperability Handling
    • Funding: transnational University Limburg
    • PhD. student: NN
    • Other researchers involved: Nico Roos
    • Period: March 2010 – March 2014
    • Short description: Interoperability is an important problem in open multi-agent systems and in electronic data exchange. Example of application domains in which the interoperability problem arises are: bioinformatics, supply chain Management and e-commerce. The networks in which collaboration takes place using Internet technology will be far more dynamic than traditional forms of collaboration. As a result the development of standard ontologies meeting the needs of all parties between which collaboration might occur, will be almost impossible in many areas and will certainly lag behind the needs. Hence, the development of tools that establishing a mapping between ontologies on demand, is crucial. This research project aims at developing such tools along four lines. First, extending the domain independent methods for learning a solution for the interoperability problem. Second, deriving a solution from meta-level descriptions of the ontologies. Third, deriving a solution from imprecise descriptions of concepts making up the ontologies. Fourth, handling data conflict resulting from the use of different representations using general domain knowledge.
  • Bee-Swarming in Robot Collectives (Bee-puck)
    • Funding: NWO and Maastricht University
    • PhD student: To be appointed
    • Postdoc: dr. Steven de Jong (starting June 2010)
    • Other researchers involved: K. Tuyls; G. Weiss; D. Hennes; N. Lemmens; M. Ponsen; M. Kaisers
    • Period: September June 2010 – June 2013
    • Short description: This research project investigates how a new bee-inspired swarm intelligence algorithm, i.e. Stigmergic Landmark Foraging (SLF), can be directly deployed on a large swarm of robots. The algorithm has been shown to outperform the state of the art in simulation. This project researches how well the algorithm is capable to coordinate a large collective of robots (up to 120 ) in a situated foraging task. Robustness, efficiency and scalability will be tested. There is a variety of potential applications, such as security patrolling, monitoring of environments, exploration of hazardous environments, search and rescue in crisis management situations and others.
  • Biology-inspired Robots and Agents for Resource Management and Logistics
    • Funding: UM ‘Breedtestrategie’ programme. Part of this research was carried out under the European research grant FP6–013569 NiSIS (Nature-inspired Smart Information Systems)
    • Period: March, 2005 to March, 2008
    • Short description: The project aimed at developing techniques for handling resource management and logistic problems using agents and robots. Research focuses was on two topics: resource management and resource distribution. Resource management was studied in the domain of resource management games such as SimCity or Virtual U. Resource management games are defined as interactive computer games where the main problem for the player is to use limited resources to construct and maintain a complex virtual environment. Resource management games are used not only for entertainment, but also for educative purposes. The research focus was on on combining traditional planning techniques with adaptive learning of search heuristics in order to efficiently and effectively manage resources in realistic, challenging, dynamic and non-deterministic environments such as resource management games. Distribution of resources is a problem found in many domains, for example public transport, hospitals and factories. Obviously, there are two parties involved here, viz. a set of transport agents (such as a taxi, train, nurse, or robot) and a set of transportables (such as passengers, patients or goods). Especially the behavior of idle transport agents is of high importance for a satisfactory and sensible performance in this domain. In this project, the focus was on techniques such as swarm intelligence, potential theory and classical planning methods in order to develop a general framework for resource distribution that is both scalable and robust.
  • Distributed Model-Based Diagnosis and Repair (DMBDR)
    • Funding: This research was supported by the Technology Foundation STW, applied science division of NWO and the technology programme of the Ministry of Economic Affairs (the Netherlands). Project DIT5780: Distributed Model Based Diagnosis and Repair.
    • Partners: The project was a collaboration between University of Utrecht, Delft University of Technology, Maastricht University and the Dutch National Aerospace Laboratory (NLR).
    • Short description: The goal was to develop techniques for diagnosing multi agent planning systems, partly based on existing model based diagnosis methods. The application domain used as a test bed is arrival, departure, taxi and gate planning of an airport. At Maastricht University, the focus was on the execution phase of a plan. During plan execution, agents should monitor whether the plan is still conflict-free. For this purpose, we developed a model which agents use as internal representation to detect conflicts at an early stage.
  • Medical Information Agents (MIA)
    • Partners: The MIA project was a collaboration between the Amsterdam Academic Medical Center, Maastricht University, and the Netherlands National Research Institute for Mathematics and Computer Science.
    • Short description: The project focused on how software agents can aid hospital workers. The goal was to provide agent support for three important directions: (a) automatic feedback to actions that are not in line with clinical-practice guidelines, (b) automatic retrieval of medical literature, and (c) scheduling patient treatment. These directions are addressed in three subprojects, each with its own agent. Each of the agents may improve its behaviour by taking into account results of the other two agents.