Department of Systems Theory and Design

The intelligent agents' research team of the Department of System theory and Design (DSTD) lead by prof. Janis Grundspenkis and assistant prof. Egons Lavendelis is working in the rapidly developing branch of artificial intelligence systems – intelligent agents and multi-agent systems. The theoretical research includes the following directions. Agent interaction protocols for different purposes, for example, for finding the best contracts based on multiple criteria and optimal task allocations in the multiple task allocation domains have been studied. Combination of the concept of the autonomous agents and architecture has been studied in terms of the holonic multi-agent systems to reduce the complexity of interactions among agents. The proposed architectures allow implementing open systems, whose functionality can be extended without changing existing code. Another theoretical direction of the research is agent oriented software engineering. Both new techniques for certain development steps and the whole agent-oriented software engineering methodology have been developed. Additionally a CASE tool supporting the development methodology has been implemented. The research team has applied the results of theoretical research in practical applications of the following domains: intelligent tutoring and knowledge assessment systems, mobile applications, knowledge management systems, insurance information systems and management systems for multiple robots. The developed ideas and approaches have found their further advancements in several CHIST-ERA (FP7 Eranet initiative) applications currently pending the results announcement, H2020 ICT call 1 proposals and industrial joint establishments. The following main results have been achieved in the intelligent agents and multi-agent system research: 1. Agent based intelligent tutoring system development methodology MASITS that combines the most suitable agent oriented software engineering techniques with the domain knowledge of the intelligent tutoring system development. 2. MASITS tool that is a CASE tool supporting and partially automating the intelligent tutoring system development process with MASITS methodology. 3. Three multi-agent architectures have been defined. The first one is open multi-agent architecture for agent based intelligent tutoring systems, the second one is for multi-robot management system, while the third architecture is for agent based intelligent tutoring system development. The architectures reduce the complexity of agent interactions and create open systems. 4. Mobile agents based knowledge management solution for mobile devices. 5. Multi-criteria auction mechanism for finding the best proposal based on various criteria. The mechanism has been applied in the logistics marketplace prototype. 6. Task allocation protocols and first prototype for the task allocation in the multi robot system. 3 Most important projects and …