Abstract

This modelling study will analyse the reaction of cars travelling along a straight road to emergency vehicles. Cars will be modelled to travel along a straight road with dependencies to other cars and the road itself. The user will then be able to introduce an emergency vehicle into the road system to which cars will respond to by moving if appropriate to allow the emergency vehicle to overtake.

The aim of this modelling activity is an initial study of this complex and unpredictable activity of how cars can respond to emergency vehicles when the space into which the cars can move on the road is, by definition of the road width and other vehicles’ positions, severely restricted. It is not expected or even necessarily considered possible to be able to accurately define the response of all cars in all such situations because all situations are different and any response is very much affected by last minute decisions of the drivers. However this study will provide some initial investigation into this phenomenon which could at a later date be expanded upon to form part of a more complete and accurate model of road based systems, for use perhaps in virtual reality driving simulators or as part of a driver training system.

This study will commence with an LSD specification of the system allowing each agent to be clearly defined (to a limited level of complexity) with dependencies and observables. This will then be modelled in practice by building a model of the system in tkEden. The model will not be constructed in a way that attempts to create the full functionality of the LSD specification from the start. Rather the system will initially be designed very simply, beginning with cars and a road simply existing together. Dependencies between cars and the road and cars and other cars will then be added to the model, and a clock developed to allow vehicles to move. Once a suitable level of complexity has been accomplished the emergency vehicle agent and related dependencies and observables will be specified and added visually to the model.

It is expected that some initial assumptions about the ways in which cars should respond to other vehicles will later be modified as a result of feedback provided by early attempts at modelling this system. Therefore the final Eden model may have a slightly different specification to the original LSD one. It may also be that this problem is considered too large to complete fully. Indeed the definition of ‘a fully complete model’ would be hard to give as further definitions (closing the model to match all the real world agents, dependencies and observables) would be very large in quantity.