Path flows are indispensable sources of traffic demand information for traffic control. In particular, for real-time traffic control, a dynamic path flow is the most important information because current condition of the road network is depicted by using this dynamic path flow. Traffic counts measured by vehicle detectors or vehicle sensors directly reflect the time variations of traffic flows and are applicable for on-line estimation of traffic demand. Therefore, a model is proposed for recursive estimation of dynamic path flows from traffic counts. This model is formulated based on dynamic relation between traffic counts and path flows. An algorithm to estimate path flows recursively is developed. Experiments with simulated data suggest that the proposed model is useful to estimate dynamic path flows. Moreover, time variations of traffic flows can be provided for real-time traffic control by the proposed model. (A*)
Abstract