Back to search

National Technical University of Athens (NTUA)

Inference of transportation networks from sparse tracking data

Abstract

dc:description

The commoditization of tracking technology, e.g., smart-phone applications involving check-ins, real-time navigation applications, fleet management, etc., provides us with a wealth of tracking data that, when utilized properly, will allow us to derive road and transportation networks. The research challenges in the form of map inference methods have been addressed to a limited extent in literature. Existing methods are characterized by limited geographical scope, small-scale tracking datasets, and unconvincing map construction results. Thus, sophisticated map inference algorithms are needed to improve over the current shortcomings and provide methods that can also be used in a practical setting.The present thesis contributes to this knowledge by proposing automatic transportation map inference algorithms for the simpler case of road networks, and the more complex case of semantically more expressive multimodal networks-of-interest. As a result of our study of automatic road network inference, we propose the TraceBundle algorithm, which extracts an arbitrary road network from large vehicle tracking datasets. A second method, the TraceConflation algorithm addresses noisy and sparse datasets. This method allows for the updates of existing road maps, both in terms of geometry and connectivity. The merit of these techniques is the automatic inference of navigable road networks of high spatial accuracy with respect to their geometry, enriched with additional attributes such as permitted maneuvers and road categories.Besides GPS tracking data and road networks, this thesis also studies a novel technique for dealing with user generated geospatial tracking data derived from social media applications. The proposed method, i.e. the Network of Interest algorithm, allows us to discover transportation hubs and critical transportation infrastructure from geocoded tweets. To further motivate and facilitate researchers and practitioners working in this area, we have created the http://www.mapconstruction.org which is an online repository containing source code of the state-of-the-art algorithms, as well as datasets for testing and evaluation.Finally, to investigate and demonstrate the applicability of our map inference algorithms in additional application domains, the map-construction approach has been applied to the visualization of eye tracking data.

Degree

thesis:*
Grantor dc:publisher
National Technical University of Athens (NTUA)
Year dc:date
2014

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Karagiorgou, Sofia
  • Καραγιώργου, Σοφία

Subjects

dc:subject × 14

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/41510

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Karagiorgou, Sofia; Καραγιώργου, Σοφία. Inference of transportation networks from sparse tracking data. National Technical University of Athens (NTUA), 2014. http://hdl.handle.net/10442/hedi/41510