Project: Multi-Objective Signal Control for Urban Enviroments

Reference: STP 14/6/44

Last update: 19/04/2013 14:22:04


Traffic signal control is increasingly seen as a way to implement social policy and not just a way to improve traffic safety and efficiency. Local authorities expect it to deliver priority for public transport, easier and safer walking and cycling, secure access for the mobility impaired, cleaner air and, more generally, revitalised town centres. The objectives have therefore become multi-dimensional and the scope of control multi-modal. This calls for innovative solutions. To help fulfil the expectations of local authorities, a software tool for multi-objective traffic signal optimisation will be developed, using fuzzy logic for the control and a genetic algorithm for the optimisation. Innovtive aspects of the work relate to; the expression of junction-specific objectives in natuarl language, the use of fuzzy logic for multi-modal control, multi-objective optimisation taking user responses into account, the consideration of time dependency iun modelling user reactions, and the inclusion of the mobility impaired. To ensure compatibility with existing systems, the proposers will liaise with with the Urban Traffic Management and Control (UTMC) Development Group.


The project will develope a software tool for multi-objective signal optimisation. This will use fuzzy logic for the control and a genetic algorthm for the optimisation. The optimisation will allow local authorities to use traffic signal control to deliver alternative priorities such as priority for public transport, easier and safer walking or access for the mobility impaired. The prototype tool will be evaluated on a test site and a report produced about its efficacy and usefulness for practioners.


Imperial College of Science Technology and Medicine
Centre for Transport Studies, Department of Civil and Environmental Engineering, London, SW7 2BU
0207 594 6089

Contract details

Cost to the Department: £113,167.00

Actual start date: 01 January 2003

Actual completion date: 31 December 2007


Multi-Objective Signal Control for Urban Environments
Author: Prof. Michael G H Bell, Sonal Ahuja and Dr Jan-Dirk Schmoecker
Publication date: 01/03/2007

Summary of results

  1. There has been a gradual shift in thinking, most recently captured in the Manual for Streets, about the function of streets. This may be characterised as a movement away from vehicles to people. Most attention to date has been given to street design, so the MOSCUE project constitutes an attempt to extend this new thinking into the realm of traffic signal control. The aim of the project was to develop a form of traffic signal control that would allow the traffic engineer much greater freedom to set priorities locally to suit the intended function of a particular junction, while not causing unintended disruption at the network level. The solution chosen involves the use of vehicle and pedestrian sensors together with fuzzy logic for on-line traffic signal control, the setting of acceptability thresholds for different aspects of junction performance (referred to as KPIs), and the use of microscopic vehicle and pedestrian simulation to optimise off-line the fuzzy set membership functions used in the on-line control.

    A multi-objective traffic signal controller (MOSCUE) has been developed and tested in simulation. The signal control uses fuzzy logic where the fuzzy set membership functions are optimised according to the Bellman-Zadeh principle of fuzzy decision-making. This approach is both practical for the traffic engineer and efficient, as it leads directly to a signal controller which delivers Pareto-optimal performance with respect to the objectives. Signal control priorities depend ultimately on political decisions about the primary function of the junction. Therefore MOSCUE allows the traffic engineer to prioritise objectives easily by setting acceptability and unacceptability thresholds for each KPI. Particular attention in the project is given to pedestrians. The membership functions for the fuzzy sets are optimised by a genetic algorithm coupled to the VISSIM microscopic vehicle and pedestrian simulator. The scope of the simulation encompasses not only all road user groups but also neighbouring junctions so that the optimisation can consider impacts at the network level. MOSCUE has been tested in simulation for the Marylebone Rd - Baker Street intersection in London at which pedestrian as well as vehicle flows are high; further tests are underway for junctions in Birmingham. Test results demonstrate the feasibility of MOSCUE and illustrate the sorts of trade-offs implied by a more pedestrian friendly form of signal control strategy.

    MOSCUE fits within the MAINSPRING (Management of Integrated transport Networks - Structured Programme of Research Into the Next Generation) initiative.