Microscopic modelling of urban traffic flow has traditionally focused attention on the dynamics of individual vehicles and drivers' perception of gaps to neighbouring vehicles. The introduction of an Integrated Road Traffic Environment presents a new challenge to the modeller: superimposed on simple car-following behaviour will be drivers' reactions to information communicated via RTI (Road Transport Informatics) devices. A formalism is needed for representing complex behaviours which is flexible, lucid and modular. The challenge is compounded with the dearth of behavioural theories to date (although some may arise from other DRIVE projects). Therefore, the author first introduces a few generic data structures to enable description of the variety of information that can be communicated via RTI devices. With this simplification, it is seen that some simple behavioural rules may be stated in English. The formal rule language is then introduced. It is shown, via examples, that rules can be used to state the specific adjustments to traffic parameters (such as route, vehicle gap, perceived costs at a junction etc.) that should be enacted given specific traffic conditions and RTI information. The language offers the benefits of being clear and unambiguous whilst being machine readable. One objection to representing theories with use of rules is that large collections of these become unmanageable. However, it is shown that rules can be integrated into a 'decision tree' structure which not only reveals the impact of adding and deleting rules but also provides an alternative graphic description of theories. Efforts to exploit and extend the rule formalism in modelling are briefly described.
Abstract