A typical daily urban traffic includes private cars with one driver and rarely one or two passengers in them. This is most likely due to people following a daily routine to get their jobs sometime early in the morning and get back home sometime later in the evening. This is advantageous in the sense of independence of individuals, i$.$e$.$ choose their times, their routes, drive their own car etc. However, the consequences on the environment, individual high costs in fuel consumption, cost for parking space etc. is the price to pay for the freedom of personal choices.

Statistical data show there are a lot of people moving daily towards the same direction and approximately at the same times. Therefore, $$sharing$$ might be a possible option. For example, a person, A$,$ drives to work from location X $($home$)$ to a location Y $($workplace$)$ every morning at $9$am$.$ Another person, B$,$ drives to work from a location close to X $($home$)$ to a location close to Y or passing by Y $($workplace$)$ every morning at around $9$am$.$ It sounds reasonable for A and B to share a drive thus cutting the costs in half. The main issue that this does not happen is that people A and B do not know that they have similar routines and they do not know each other.

If those people were assisted by a suitable mobile application then they could have considered sharing. The app would recognise daily routines of individuals $($locations$,$ times, etc$),$ record any personal preferences $($types of cars, capacity, passengers availability etc.$)$ and by communicating with other individuals could recommend sharing. Each app need not be monolithic in the sense that it can get data and knowledge from other relevant apps $($e$.$g$.$ maps, location, time etc.$).$ In addition, the whole system may consist of a set of apps of different individuals which share personal data and knowledge in order to match the daily routines and suggest sharing. Issues like trust, safety, cost sharing etc. may be taken into account too.

One could argue that the solution to this problem can be facilitated by using AI technology and, more specifically, a Multi$-$Agent System $($MAS$).$ You are requested to propose$/$design a MAS which aims to facilitate sharing of driving to a destination between people who have similar daily moving routines.

You can assume that each individual person possesses a mobile phone and has access to telecommunication services, such as the internet, location based$-$services etc.$)$ at all times.

You are required to submit a structured typewritten report of no more than $10$ pages $($including any diagrams, figures and references as appropriate$).$ The report should support the thesis $($proposal for a MAS$)$ and should address the issues in appropriate detail that includes background theory, justification of choices and examples where necessary.

Issues that you should address in your report $($not ordered in any way of preference or presentation$)$:

Suitability of the problem to be resolved by a MAS and its characteristics

Suitability and type of the environment in which the MAS will operate

Any other assumptions about the environment, the external services and sensors available

Justification of why the proposed software matches the characteristics of agents$/$agent systems

Types or proposed agents and their structure $($abstract architecture$)$

Abilities $($abstract specification of actions$)$ for each agent

Knowledge and Reasoning of each type of agent in the MAS

Knowledge sharing between agents

Types of interaction

Organisation of the MAS, collaboration and$/$or competition

Communication between agents

Interaction protocols

Negotiation $($if necessary$)$

Any other challenges in the design of the MAS

A straight short answer to any of the above is not enough. You are strongly advised, for each one of the above points that you are to address in the report, to revisit the relevant lecture notes and possibly other sources, in order to identify exactly what your answers should include. Make sure you accompany your answers$/$suggestions with the relevant background, definitions, analysis, justification, etc. in order to demonstrate your understanding and skills gained through this module.

You are NOT supposed to produce any code or partial implementation of agents. You are also NOT supposed to focus on the interface of the front$-$end of the app.