Dr. Vangelis Kourakos Mavromichalis

Department of Information and Communications Systems Engineering
University of the Aegean
Greece


Research
Publications
Short CV

Programming in Logic

Contact Information and Office Hours






Research

ICAGENT: Towards a Generic Framework for the development of Intelligent Collaborative Agents.

In complex, resource bounded environments, agents face inherent limitations with respect to their perception, reasoning, performance and cooperation abilities: They cannot be fully aware of the changes that occur in their physical environment; the available time to compute responses is always limited and bounded to the time that they have until the resources get exhausted; they have to plan and to achieve their goals without exceeding environmental and own resources, and finally, they often need to collaborate with others to achieve better results.

Aim: To develop intelligent agents for dynamic and unpredictable environments capable to plan and act effectively individually or in collaboration with other agents (including humans).

To overcome the above limitations, a resource bounded agent must be able to plan and act deliberatively with respect to its own, others’, and environment resources, adapting to newly detected events and facts, either individually or in collaboration with others. On the other hand, an agent needs to react to unforeseen events/facts that provide opportunities for achieving its goals or to events/facts that may seriously affect its or the collaborators' mission. In either case, an agent may perform individually or in conjunction with other agents.

Objective: The main objective is the establishment of a generic framework (ICAGENT Framework), for developing intelligent agents capable to adapt their behaviour, to collaborate with other agents (including humans).

The ICAGENT Architecture.

In ICAGENT framework using the BDI-model of agency, resource bounded agents are able to:

  1. Reason practically, that is to decide what state of affairs to achieve and how to achieve these with respect to:

    • Their own mental state (beliefs, desires, intentions, context of action).

    • Other agents mental states.

    • The knowledge that they have about the state of the physical environment.
       

  2. Collaborate with other agents, forming shared plans and goals based on the collaborative model of activity of SharedPlans (Grosz 1996).
     

  3. Adapt their behaviour, balancing between:

    • Reactive behaviour.

    • Deliberative behaviour.

    • Social behaviour.
       

Key issues: Towards this objective key issues are the following:

  • Provide the adequate reasoning tasks that enable agent to monitor the environment, recognize the situations that must take the initiative to act, to elaborate its plans hierarchically, to reconcile its plans forming new intentions and finally to realize its plans by performing the intended action. All these reasoning tasks are performed in parallel and communicated only with the knowledge base, enable the agent to plan and realize its plans simultaneously and coherently.

  • Provide clear definitions of deliberative and reactive (individual as well as cooperative) planning in terms of agent's reasoning tasks and mental state.
    In complex environments an agent may need to adapt its behaviour several times in response to changes in the environment while it pursues a specific goal. In ICAGENT framework, such an adaptation is considered as a “tuning” process, with respect to the way several mental actions are performed. In a greater extend than existing frameworks based on layered architectures; ICAGENT relates agent’s flexible behaviour to cognition and sociability, supporting the management of plans constructed by the agent’s mental and domain actions in a coordinated manner. Therefore, the type of behaviour adopted by an agent at a specific time point is considered as a property that emerges as the agent performs, according to the perceived state of the environment, the occurring events, and to the agent’s mental state.

  • Provide the necessary mental attitudes and mental actions for collaborating with other agents forming shared goals and plans. ICAGENT exploits these practical-reasoning tasks to support the implementation of the SharedPlans model of collaborative activity.
     

Developing Intelligent Interface Agents with the ICAGENT Development Framework.

To adapt information effectively to instructional contexts, we need active learning environments that reason about the domain, users' mental state and information needs, the general context of action, as well as about strategies and methods for performing domain actions on behalf of the user and designing effective information presentations in the course of dialogues, using multiple media.

In this context, we investigate the development of intelligent collaborative interface agents that act as tutors in the context of software applications. Utilizing the ICAGENT  framework we build interface agents that deliberate socially with users following the SharedPlans model of collaborative activity (Grosz 1996). Social deliberation requires interface agents to make their desires and intentions clear to the application users, to understand the context of their activity and to reconcile their own and users’ desires in the overall context of action. Being in constant communication is important for interface agents to exhibit robust and trustworthy behavior, while understanding the context of activity and reconciling users’ desires allows agents to recognize the situations where users need help.

ICAGENT provides generic mechanisms for plan management, for the recognition of actions and the reconciliation of collaborating parties’ desires and intentions, for planning and acting towards achieving domain goals and for recovering from mistakes by initiating collaboration. The collaborative interface agents are able to:

  • Form hypotheses about and recognize the desires of users.

  • Detect conflicts during the course of activity, exhibit helpful behavior and initiate collaboration.

  • Deliberate socially with the users and act jointly with them towards achieving shared goals.

  • Advise and/or assist users in a trustworthy way.

 

Publications

Short CV

Vangelis Kourakos Mavromichalis completed his undergraduate studies in the University of Aegean Department of Mathematics, Greece. He received a Ph.D degree in Artificial Intelligence from the Department of Information and Communications Systems Engineering of the University of Aegean (UoA) in 2006. Currently, he is working as technical staff at AI Laboratory in the Department of Information and Communication Systems Engineering of the University of Aegean. Also, he is member of the following research groups and societies:

Main Research Interests

  • BDI Agents
  • Collaborative Agents
  • Intelligent Multimedia Systems

General Research Interests

  • Cognitive Science
  • Cooperative Planning
  • Rhetorical Structure Theory
  • Non-Monotonic Temporal Logic

Contact Information

Dr. Vangelis Kourakos Mavromichalis
Dept. of Information and Communication Systems
University of the Aegean
Karlovassi 83200 - Samos
Greece
Email:

Office Hours

09:00am - 03:00pm