Organization responsible: Thales Research & Technology United Kingdom

People involved:

• Tim Masterton (coordinator, project leader, researcher)
• Andrew Batchelor (researcher)
• Diven Topiwala (until 31-12-2008)

Project description:

The project aims to investigate tasking mechanisms in groups of autonomous agents. An important aspect of CDM is the efficient organisation of tasking of the available resources or actors (these could be humans or some kind of agent or a combination of both). In particular, if a number of actors are to be collectively used in the pursuit of a given common objective, it is vital to ensure that the available actors are tasked in a manner that is as close to optimum as possible.

A core factor in the problem of tasking and re-tasking in dynamic environments is the architecture implemented to facilitate rapid and dynamic tasking. TRT (UK) proposed, and began prototyping, a distributed tasking architecture based in agent technology.

The research in this project aims to answer to the following questions:

• How do learning techniques compare to empirical techniques for merging and prioritising the overall requirements of the generic CDM system?
• Which are the best techniques for merging and prioritising the overall requirements?
• How do learning techniques compare to empirical techniques for scheduling of the overall goals of the system under physical constraints?
• Which are the best techniques for scheduling of the overall goals of the systems under physical constraints?
• Which techniques are best for task to actor allocation?
• Which techniques are best for re-scheduling optimisation, whilst prevent over adaptation?
• Which techniques should be used for recovery from communication failure or actor deletion?
• Which of these techniques work best in real time?

Publications: