The Prometheus Methodology

The Prometheus methodology is a detailed process for specifying, designing, and implementing intelligent agent systems which we have been developing over the last several years. Our goal in developing Prometheus is to have a process with defined deliverables which can be taught to industry practitioners and undergraduate students who do not have a background in agents and which they can use to develop intelligent agent systems.

Prometheus distinguishes itself from other methodologies by supporting the development of intelligent agents, providing "start-to-end" support, having evolved out of practical industrial and pedagogical experience, having been used in both industry and academia, and, above all, in being detailed and complete. Prometheus is also amenable to tool support and provides scope for cross checking between designs.

The methodology consists of three phases: system specification, architectural design, and detailed design. An overview of the methodology, including its phases, deliverables, and intermediate products, is depicted below. Although the phases are described in a sequential fashion it is acknowledged that like most Software Engineering methodologies, practice involves revisiting earlier phases as one works out the details. 

For more details see the book Developing Intelligent Agent Systems: A Practical Guide and also the relevant publications below.

Prometheus Design Tool (PDT)


The Prometheus Design Tool (PDT) is a tool that supports the Prometheus methodology for building agent based systems. PDT is developed and supported by the RMIT Intelligent Agents Group. For more information on PDT including downloading the tool click here.


Testing Agent Systems

The use of agent technology for building complex systems is increasing, and there are compelling reasons to use this technology. Autonomy, flexibility in terms of being able to achieve a particular task in a number of ways, and robustness in terms of failure recovery, are some of the key features of such systems. However the complexity of the systems that can be built using this technology, does create concerns about how to verify and validate their correctness. 

Unit Testing

This paper describes an approach to unit testing of plan based agent systems, with a focus on automated generation and execution of test cases. Design artefacts, supplemented with some additional data, provide the basis for specifica- tion of a comprehensive suite of test cases. Correctness of execution is evaluated against a design model, and a comprehensive report of errors and warnings is pro- vided to the user.


Verification of Design Models