Multi-agent systems are a central topic in modern mainstream artificial intelligence and not adequately covered by any other unit. This is a major gap for a contemporary AI course.

There is a substantial amount of research on MAS or involving MAS going on in the faculty that spans a diverse range of domains, including collective behaviour, optimisation, energy and other fields. These research directions currently have no adequate representation in our courses and can thus not train up future research students.

The proposed unit will fill this gap.

This truly interdisciplinary unit will also serve as an elective for other Masters courses.

On successful completion of this unit, you should be able to:

1. judge whether a particular real-world problem or application can usefully be modelled as a MAS
2. select MAS modelling methods suited to the problem
3. apply formal MAS modelling approaches as appropriate
4. implement and deploy MAS models to answer the relevant questions about a given real-world scenario
5. discuss the limitations of the modelling approaches and use multi-model solutions to mitigate or overcome these

020119

A multi-agent system (MAS) consists of a number of autonomous agents interacting with each other and with their environment. MAS is one of the fastest-growing areas of AI and a very general paradigm to understand many complex natural phenomena, such as the behaviour of ant colonies, fish swarms and human groups. Conversely, MAS approaches are crucial in the design of some of the most cutting-edge AI and cyber-physical systems, such as swarm robots. Hybrid cyber-physical systems, in which natural and artificial agents interact, represent the third important form of MAS. The internet, where millions of humans and computational agents interact in intricate and complex ways, is the paradigmatic example of such a hybrid.

This highly interdisciplinary unit discusses the most important methods to describe, analyse and design MAS and discusses their practical applications in scientific modelling and artificial intelligence.

On-campus

Examination (2 hours and 10 minutes): 40%; In-semester assessment: 60%

Minimum total expected workload equals 12 hours per week comprising:

(a.) Contact hours for on-campus students:

• Two hours lectures
• Two hours laboratories

• A minimum of 8 hours of personal study per week in order to satisfy the reading tutorial, practical and assignment expectations.