Dynamic travel demand estimation and prediction methods

Availability of accurate trip demand estimates plays a key role for both long term and short term traffic planning as well as for on-line applications of intelligent management strategies that require reliable forecasts of traffic demand so that users’ response to varying flow conditions, both observed and communicated, in different locations and in different time intervals, can be properly taken into account and anticipated. To guarantee consistency with respect to temporal and spatial dimensions, traffic demand estimates and predictions must reflect both time variability and network patterns. This calls for a problem known in literature as Dynamic Demand Estimation problem (DDEP); it can be formulated as an off-line problem for medium to long term planning and design, or as an on-line problem for real time ATMS/ATIS applications. In both cases, traffic dynamics are dealt with at network level. Nowadays, demand data derive from advanced traffic surveillance systems based on the collection of several heterogeneous traffic measurements both in fixed locations and over specific corridors or paths. Such recent technology developments suggest new promising and challenging chances, not fully addressed yet. In this chapter, the main formulations of the demand estimation and prediction problem are described first, along with the related literature review for both the off-line and the on-line approach; then, some relevant study-cases are illustrated; finally, future perspectives relying on new emerging technological opportunities are envisaged.