The problem of Short-Term Traffic Forecasting consists in estimating the number of vehicles that will use a specific route within a city in the near future.
In the project PLUG-IN our research group developed the system ITACA (Intelligent TrAffic ForeCAster), an artificial intelligence tool based on fuzzy clustering and committees of neural networks able to forecast traffic in the short term using data collected from multiple sources. The forecasting horizon was from 5 to 10 minutes in the future.
Traffic flow measurements can be obtained through various sensors, most commonly inductive loops or cameras. Data is collected in a data center or on a cloud and the traffic flow forecasts obtained from ITACA can be made available to the traffic manager for real-time decision making, and to end users through panels and / or smartphones.
Current urban traffic forecasting tools use flow analytic models of the vehicular flow based on the analogy with the dynamics of liquids, or microscopic models that simulate the vehicles individually. However, these types of modeling are not able to face real-life road networks in real time, and fail when exceptional events occur. ITACA is instead able to scale on complex urban networks.
The availability of an efficient traffic forecaster like ITACA within a system for urban traffic management can allow the local administration to decongest traffic and reduce travel time, resulting in significant energy savings, reduction of pollution and a better quality of life for the population (given the improvement of health and the stress reduction).