Evanthia Kazagli

The Royal Institute of Technology (KTH), Stockholm, Sweden

August 14, 2012, 11:00, Room GC B3 424 (click here for the map)

Estimation of arterial travel time distributions from automatic number plate recognition data using mixture models

<p>Automatic Number Plate Recognition (ANPR) data have been widely used for estimation of travel time and travel time distributions, mainly in the case of freeways. The objective of this work is to formulate a finite mixture model for the estimation of arterial travel time distributions based on ANPR data. A black spot when extracting arterial travel times from ANPR data concerns vehicles that do not traverse the monitored section directly, but stop in between for various reasons (loading/ unloading, buses stopping at bus stops etc), resulting in higher than the usual travel times (invalid observations). Assuming that the population of ANPR travel times is generated by two different subpopulations (components) -one deriving from non-stopped (valid) vehicles and one from the stopped- finite mixture models can be used at a first level as clustering technique to separate these two components. In an attempt to reinforce the model, explanatory variables such as weather conditions are included in the estimation. In arterial networks, route travel times are likely to vary among the valid observations, even over small time intervals. This is due to such factors as traffic lights, buses stopping, vehicles turning mid-link delaying following vehicles, etc (Robinson, 2005), resulting in a multimodal probability density function of travel time. In this context mixture models can be used - at a second level - for the estimation of route travel time distribution. A very important aspect is the assumption for the underlying distribution. The common assumption of normal distribution of travel time is replaced by log-normal (in this case mixture of log-normal distributions). </p><p> Robinson, S. (2005). The development and application of an urban link travel time model using data derived from inductive loop detectors, PhD Thesis, Centre for Transport Studies, Department of Civil and Environmental Engineering, Imperial College London, United Kingdom. </p>


Evanthia Kazagli holds a Diploma of Rural and Surveying Engineering from the National Technical University of Athens (NTUA). She is currently a Master student in the division of Traffic and Logistics at the Royal Institute of Technology (KTH), where she is elaborating her thesis under the supervision of Prof. Haris Koutsopoulos. Her work deals with the estimation of arterial travel time distributions. Her research interests include among others intelligent transportation systems and travel time estimation and reliability.