The rule-based system CLAIRE was designed to enhance the ability of existing urban traffic control systems to cope with congested situations. This paper reports how a preliminary assessment of this capability was undertaken. It was considered essential for the evaluation environment to be controllable and repeatable. To achieve this, the graphical micro-simulation package AIMSUN was used to model vehicle behaviour through a set of signalised junctions, with simulated detectors providing information to CLAIRE. The model was based on part of the city of Leicester, operating under the SCOOT traffic control system, and it was calibrated using data obtained from on-street surveys. The evaluation environment was in near real-time, and consisted of two interconnected VAX work-stations, with interfacing modules to translate between CLAIRE, AIMSUN and TRANSYT. Five evaluation runs were carried out; two compared CLAIRE's effect with that achieved by re-optimising with TRANSYT, and three compared it with an approximation to SCOOT control. The evaluation showed that CLAIRE could improve upon poor fixed-time plans; whilst improvements with respect to SCOOT were only slight, this is thought to be due to the relatively low level of congestion in Leicester. The paper concludes with recommendations for further trials. (A)
Abstract