Development of network-wide traffic signal control algorithms to achieve stable network conditions for over-saturated traffic conditions

Traffic flows through an urban area are determined by drivers’ desires in terms of origin and destination and their choice of routes in response to perceived costs of using alternative routes. These costs can be thought of as a combination of time and distance with other influences such as restricted knowledge of the network and user preferences for certain route attributes e.g. avoid right turns or give-ways as much as possible. These costs are influenced by the flow on each route and in general the only way in which a traffic engineer can affect the costs is by the introduction of physical measures such as traffic calming or new capacity or by using the signalised junctions within an area to directly change the costs of using certain routes. This research is concerned with the use of responsive signal control systems and their use to optimise the traffic flows within a network.

Traffic signals have long been used at operational level as a tool to manage the conflicts in and balance the delays to traffic streams at individual intersections as well as at a network level. In contrast to this operational role, traffic signals have been recognised to have a strategic role in that they affect drivers’ long-term route choice in a network. Although there is much practical evidence at the operational level and some research experience at the strategic level, these two functions of traffic signals have to date not been integrated.

The aim of this research project is to develop network-wide traffic signal control systems which take into account of day-to-day variability (in demand and supply), of driver learning/adaptation, and of multi-class traffic flow conditions, and which produce stable network conditions and reduced in CO2 emissions in over-saturated conditions.

During the course of the research interactions and constraints on the sub-components of the algorithms forming the Urban Traffic Control framework will be investigated. These include the predictive model performance and implications of feedback errors, calibration issues, implementation of new objective functions, the effect of constraints and possibilities of including demand management as a strategic policy.