Organization responsible: TUD, ARS T&T

People involved:

• Hans van Lint (promotor, project leader)
• Ou Qing (PhD student)
• Henk van Zuylen
• Fred Zijderhand
• Dusica Joksimovic

Project description:

This project aims at developing the tools and methodology for estimating the prevailing traffic conditions in a heterogeneous (urban, provincial and freeway) traffic network. For that, this proposal aims at a network-wide traffic data fusion framework based on intelligent traffic agents, in which the information available from whatever traffic sensor on whichever location (road) and time is maximized, generalized and mapped into a coherent and consistent representation of traffic conditions on an entire traffic network. The research can use the unique real-time traffic data test bed and a huge database of real traffic data which is readily available in the Regiolab-Delft project.

1. Consistent estimation of traffic density and speed. Usually, traffic density is hardly available from loop detector, while it is a crucial parameter for traffic prediction and fundamental diagrams. In this topic, we will make the best of individual travel times from floating car data or cameras combined with loop counts to estimate past and current traffic density on each segment. And this algorithm is expected to contribute to traffic prediction.
2. Validation of the algorithm by real-life data. For now being, our newly developed algorithm PISCIT has not been validated by real-life data. Neither with the algorithm as proposed above. So we are supposed to validate both algorithms by cooperating with Portland University where stacks of floating car data are available.
3. Estimation of Route Choice Algorithm.  In real-life application for network-wide data fusion, we will have to get to know the routes vehicles travel on, otherwise individual travel time information wont work properly combined with other traffic information.