The NDW Traffic Observatory
In the Netherlands, the National Data Warehouse for Traffic Information (NDW) is an organisation best known for the product that gives it its name: its enormous database of both real-time and historic traffic data. We are colllaborating with NDW to make this database intelligent and searchable, and to extend the services offered towards coupling of more data and deriving more information from it.
NDW is a unique alliance in which 19 public authorities work together, learn from each other, and consolidate their data and other resources. Its goal? To apply the right data to obtain optimal traffic management and to provide road users with the best possible information resulting in less congestion, lower emissions of CO2 and other pollutants, and improved safety.
Reliable and complete traffic information is critically important for Intelligent Transportation Systems (ITS). From personal traffic apps to traffic control centers, the efficiency and effectivity of the services delivered require an accurate and reliable estimate of the prevailing traffic state. In this project we develop a new innovative multi-scale framework that can deliver these traffic state estimates for large road traffic networks. This framework is based on a fusion engine that can take in watever data sources are available (roadside sensors, GPS equipment, geographical data, etc) and turn these data into one coherent picture of the traffic conditions. The multi-scale approach makes this project unique in the world. Our framework will produce estimates on different scales that are mutually consistent and that can be used for many existing and new applications within the ITS community and transportation in general.
In the first leg of this project we develop a new idea for (superfast) searching through traffic data bases, that will make it possible to search for traffic patterns without having to plough through 200TB of raw NDW data. To this end we develop CoSi (Congestion Search engine). With CoSi completely new research becomes possible, such as wide scale analyses of traffic patterns and the dynamics of all sort of characteristics over the entire Dutch freeway network. From DiTTLab is PhD Candidate Tin Nguyen involved.
DiTTLAB Partners: TU Delft, CGI, NDW
Publications:
- Nguyen, H. N., Krishnakumari, P., Hai, L. V. & van Lint, H (2016). Traffic congestion pattern classification using multi-class SVM. In 2016 IEEE 19th International Conference on Intelligent Transportation Systems (ITSC), pages 1059-1064. [Digital version]