CLAIRE is a European prototype Expert System for congestion monitoring, control and learning. The project has been developed in the DRIVE project V 1015, Artificial Intelligence Based Systems for Traffic Control. This paper briefly reports the main elements of research carried out by the consortium. The functions and components of the prototype expert system are described. Results of studies to identify the differences in signal control parameter definition and urban traffic policy of European cities are discussed. Identifying the differences was fundamental to ensuring that CLAIRE was compatible with, and will operate independently of, current urban traffic control systems. Through studies of congestion monitoring and measures of performance, methods to diagnose and evaluate congestion trees have been derived. The off-line modelling capability of CLAIRE, used to define remedial control strategies for congestion, based on signal optimisation and route assignment is presented. Finally the procedures developed to evaluate the success of CLAIRE at recognising congestion, diagnosing a problem and recommending the correct remedial control strategy are described. This uses a traffic simulation model. At the time of writing this paper the results of the evaluation are not available but these will be presented at the conference.
Abstract