Faculté de l'environnement naturel, architectural et construit ENAC, Programme doctoral Environnement, Institut des infrastructures, des ressources et de l'environnement ICARE (Laboratoire des voies de circulation LAVOC)

A methodology (CUPRITE) for urban network travel time estimation by integrating multisource data

Bhaskar, Ashish ; Dumont, André-Gilles (Dir.) ; Chung, Edward (Dir.)

Thèse Ecole polytechnique fédérale de Lausanne EPFL : 2009 ; no 4416.

Ajouter à la liste personnelle
    Travel time is an important network performance measure and it quantifies congestion in a manner easily understood by all transport users. In urban networks, travel time estimation is challenging due to number of reasons such as, fluctuations in traffic flow due to traffic signals, significant flow to/from mid-link sinks/sources, etc. In this research a methodology, named CUmulative plots and PRobe Integration for Travel timE estimation (CUPRITE), has been developed, tested and validated for average travel time estimation on signalized urban network. It provides exit movement specific link travel time and can be applied for route travel time estimation. The basis of CUPRITE lies in the classical analytical procedure of utilizing cumulative plots at upstream and downstream locations for estimating travel time between the two locations. The classical procedure is vulnerable to detector counting error and non conservation of flow between the two locations that induces relative deviation amongst the cumulative plots (RD). The originality of CUPRITE resides in integration of multi-source data: detector data and signal timings from different locations on the network, and probe vehicle data. First, cumulative plots are accurately estimated by integrating detector and signal timings. Thereafter, cumulative plots are integrated with probe vehicle data and RD issue is addressed. CUPRITE is tested rigorously using traffic simulation for different scenarios with different possible combinations of sink, source and detector error. The performance of the proposed methodology has been found insensitive to percentage of sink or source or detector error. For a link between two consecutive signalized intersections and during undersaturated traffic condition, the concept of virtual probe is introduced and travel time can be accurately estimated without any real probe. For oversaturated traffic condition, CUPRITE requires only few probes per estimation interval for accurate travel time estimation. CUPRITE is also validated with real data collected from number plate survey at Lucerne, Switzerland. Two tailed t-test (at 0.05 level of significance) results confirm that travel time estimates from CUPRITE are statistically equivalent to real estimates from number plate survey. The testing and validation of CUPRITE have demonstrated that it can be applied for accurate and reliable travel time estimation. The current market penetration of probe vehicle is quite low. In urban networks, availability of a large number of probes per estimation interval is rare. With limited number of probe vehicles in urban networks, CUPRITE can significantly enhance the accuracy of travel time estimation.