Unit:
Department of Informatics and TelecommunicationsΠληροφορική
Author:
TSAPELAS CHRISTOS
Supervisors info:
Δημήτρης Γουνόπουλος, Καθηγητής, τμήμα Πληροφοροικής και Τηλεπικοινωνιών, Σχολή Θετικών Επιστημών
Original Title:
Pathlet Learning for Compressing and Planning Trajectories
Translated title:
Pathlet Learning for Compressing and Planning Trajectories
Summary:
The last years, there has been a wide spread of GPS-enabled devices, which has led to the generation of unprecedented amounts of spatio-temporal trajectory data. These datasets offer a great opportunity for enhancing our understanding of human mobility patterns, thus benefiting many applications from location-based services.
In this work, we faced the problem of pathlet learning to understand shared structure in a large collection of trajectory data. The aim of this project is to extract a pathlet dictionary able to reconstruct all the input trajectories of the dataset, which can be later used as a base for route planning and travel time estimation.
Our methodology describes the process of mapping the GPS data to road segments on the map and proposes the definition of the problem as a problem of Linear Programming. Then, due to the complexity of the problem and the lack of the huge amount resources required, we implemented some preprocessing in the input trajectories which resulted in solving smaller problems. Finally, the results of our work are presented.
Main subject category:
Science
Keywords:
GPS trajectories, linear programming, pathlet learning, pathlet planning
Bsc Thesis.pdf (1 MB)
Open in new window
src.zip
6 KB
File access is restricted.
