Search this site
Search WWW

Copyright © 2001 University of Otago and its licencers. All rights reserved

Research
 

Research areas of interest:

Agent platforms.The DC&SA Research Centre has developed two platforms that provide the basic infrastructural support for agent interaction, the NZDIS platform and OPAL (the Otago Agent Platform). Opal is the one undergoing further development. We have also produced several utilities, the NZDIS Utilities, that can be used in connection with agent development -- they are available here.

Agent-Oriented Software Engineering.

 

Michael Winikoff – AOSE research:
The overall aim of my research is to find better ways of creating software. Over the past decade I have focussed on agent-based software development, looking at both software engineering aspects, and programming aspects. One of my more significant contributions to the field was the development (with Professor Lin Padgham) of the Prometheus agent-oriented software engineering methodology. I am currently interested in the following topics:

  • Concepts for designing agent software including the crucial role that is played by goals and how (and why) they should be explicitly represented in agent systems.
  • Techniques for designing and implementing flexible and robust interactions between agents, including both goal-based techniques and techniques based on commitment machines.
  • Techniques for debugging and for software maintenance of agent systems.
  • Techniques for assurance and validation of agent systems - how can we be confident that an adaptive system will behave appropriately?
  • Agent-oriented programming languages.
  • Applying social expectations to monitoring interactions (which can be software-software, software-humans, or computer-mediated human-human interaction).

Mariusz Nowostawski and Martin Purvis – Micro-agents
Our notion and implementation of micro-agents (µ-agents) support the employment of agent-based concepts at all levels of software engineering and coding. In this respect they differ from the conventional deliberative agent-based technologies, which do not support hierarchical decomposition. In addition, the use of our µ-agents does not involve a performance penalty. We have also developed MOA (µ-agents on Android), which is a port of the µ-agent technology so that it works efficiently with the Android mobile platform.

See http://www.micro-agents.net/, where the µ-agent metamodel as well as the actual µ-agent platform are described in more detail, including instructions on how to implement applications.

Agent-based Coordination & Cooperation

Tony Savarimuthu – Research into Agent-based Normative Behaviour:

  • Agent-based modeling of human behaviour (e.g. crowd behaviour in streets, fire-exit escape scenarios)
  • Modeling normative behaviour of agents in virtual societies (Second Life, MMOGs)
  • Modeling the effectiveness of social norms in creating sustainable environments (e.g. promoting recycling, energy conservation)
  • Emergence of norms and conventions
  • Emergence of signals (sanctions or rewards) which are precursors for norm formation
  • Types of norm learning (experiential, observational and communication-based learning)
  • Impact of lying on convention emergence (in the context of employing communication-based learning)

Other areas of research interests include data mining (e.g. mining norms, mining information from news paper articles), harnessing crowd-sourcing (e.g. using twitter volunteers to transform twitter posts into machine readable formats), business process modelling and location-based services.

Stephen Cranefield - Ontology representation in UML.
Agent-based Social Structural Modelling for Peace & Conflict Evolution (AbSSM4P). Given the growing interconnectedness and complexity of human societies, improved modelling techniques are essential for the understanding and prediction of social instabilities leading to conflict, insurgency, and war. The Agent-based Social Structural Modelling for Peace & Conflict Evolution (AbSSM4P) research project is a new approach for addressing this issue (see Figure 1)
Maryam Purvis & Tony Savarimuthu JBees workflow system. Adaptive and distributed agent-based workflow systems for managing and monitoring the workflow in dynamic distributed systems.

Nomadic, peer-to-peer (P2P), and wireless applications. Increasingly, handheld wireless devices are being used by people on the move. Software agent technology can be used to adapt the application to the changing environment as the user move around. The FragME2004 wireless P2P platform has been developed that supports the development of wireless games and two wireless P2P games have been developed so far: RoboJoust and Boom!.

Martin Purvis – Narrative-based Interaction. Narrative techniques are being developed to enhance the interactive nature of distributed and agent-based systems and to enrich these interactions by incorporating multimedia. See Klaas-Jan Winkel's thesis on this topic.
Security and trust in distributed systems.
Situated agents embodied as physical robots.

Distributed Computational Architectures. A new general approach for Evolvable Virtual Machines (EVMs) has been developed which has the potential to evolve hierarhchical, cooperating agent machine structures.

The Mobile Multimedia Social Networking (MMSN) project is a proposed partnership with Telecom NZ to develop technology for expanding social networking interaction opportunities across New Zealand.
Distributed Computing Technology & Applications (DCT&A) Research Project. This programme, funded by FRST, developed technology that can help organisations realise the potential of distributed interactive systems. The technical foundation of this work, software agent technology, provides mechanisms for information exchange in distributed, open systems. The research involved the development of improved agent-based development tools, supporting component services, and the demonstration of applications in the area of integrated access to distributed knowledge sources. The basic research goals of this project continue to be a focus of research at the DC&SA Research Centre. (funded project completed 2007)
Distributed Information Systems project. The goal of this work was to develop advanced distributed information system technology that enables users and enterprises to access and integrate information from all available sources in a timely and useful fashion. Agent technology has been used in the New Zealand Distributed Information Systems (NZDIS) funded by the New Zealand Foundation for Ressearch Science & Technology (FRST). (funded project completed 2002)