In-car navigation systems and traffic management centres are based on digital road network representations. In the future these systems will include dynamic travel time information. Therefore additional link attributes for dynamic routing and travel time forecasting are needed. This data can partially be generated from existing transport planning models. Matching algorithms are proposed in this paper to construct a new transport planning network incorporating precise geographical information from the detailed navigation network and the link attributes already present in the original planning network. The resulting network is optimally suited to the task of computing the current level of service. For forecasting purposes even less detailed networks are sufficient. Only motorways, major trunk roads and arterials should be represented within the forecasting networks to minimise the computational effort. The link attributes and flow dependent travel time functions have to be taken from the planning networks.
Abstract