High-quality Urban Traffic Monitoring ? Theory and Evaluation of a Promising Data Fusion Methodology
T Neumann, German Aerospace Center, Institute of Transportation Systems, DE
An efficient method to estimate queues at traffic lights is proposed. Using floating car data and providing flexible capabilities for data fusion, it yields high-quality estimates for the traffic states at traffic lights at quite realistic scenarios.
Recently, the author proposed a new statistical approach for the estimation of queue lengths at traffic lights. Conceptually based on floating car data, it provides very flexible capabilities for real data fusion which allow for the integration of nearly arbitrary additional traffic data. As a result, it is able to yield high-quality estimates for the traffic states at traffic lights even at quite low penetration rates (~2%) concerning floating cars.
The paper describes the theoretical background of the new method in detail for the first time. In this context, a common microscopic traffic flow model which is also used for urban traffic modelling, is stochastically analyzed with regard to the relation between local traffic density, inflow and queue length at traffic lights. In contrast to classical queuing theory, this allows for an integrated view on the traffic dynamics at traffic lights. Furthermore, the spatial structure of the traffic light queue can be considered explicitly.
Finally, an overview about the new method is given which also summarizes the results of a systematic evaluation campaign based on extensive simulations with about 5 million scenarios. Thereby, the extraordinary quality can be shown which is realized by the described algorithms. Very small average absolute errors (< 2 vehicle lengths) are obtained constantly even in case of very simple data fusion approaches. Furthermore, the strong correlation between real (i.e. simulated) and estimated queue lengths with correlation coefficients R > 0.975 at quite realistic penetration rates (~2%) also underlines this quality.
Association for European Transport