A Realistic Case Study for Comparison of Data Fusion and Assimilation on an Urban Network – The Archipel Platform☆

Achievements from conducted research as part of the team of the Laboratory of Traffic Engineering Transportation (LICIT) of IFSTTAR are integrated in platforms in order to develop and test new modules directly on real data. The ARCHIPEL platform focuses on data assimilation and fusion from multiple sources. The objective of ARCHIPEL is therefore to provide tools to process real-time data from different sources. This platform allows one to test and compare different approaches of assimilation and data fusion: loops, Global Positioning System (GPS), Bluetooth. Besides it offers to view details of traffic conditions in the form of space-time diagram on accurate and successive areas or larger areas by traficolor representation. In the context of ISpace&Time project, the city of Paris has been finely modelled in a traffic simulator developed by the LICIT, called SymuVia. This permits to have access to a case study in which all vehicle positions are known. In order to explore which fusion method is the most suitable depending on the context, e.g. penetration rate and data noises, these realistic data have been put in the platform database and are utilized to give a complete overview of the more relevant methodology to apply. In this work, the authors propose some fusion scenarios and present the comparison results with respect to different conditions. The final aim is to give a roadmap of the more accurate solutions, since the ground truth is exactly known.