StudiumOffene Abschlussarbeiten
On applicability of semantic place discovery algorithms for traffic regulator detection and classification

On applicability of semantic place discovery algorithms for traffic regulator detection and classification

Leitung:  Zourlidou
E-Mail:  zourlidou@ikg.uni-hannover.de
Jahr:  2019

Objective

The objective of this thesis will focus on the study of vehicle trajectories that can reveal traffic regulations through the recognition of common driving patterns (e.g. collocated events like stops, slow movements, sequences of actions, etc).  Vehicle location data and their respective motion measurements (e.g. speed) can be processed in regular intervals for extracting road rules that traffic participants must respect. In other words, by recognizing collective driving behaviour, traffic rules can be automatically mined and mapped providing up-to-date rule-aware maps. 

For the movement pattern recognition task, this thesis will explore the applicability of some known algorithms (CB-SMoT, T-OPTICS, etc) for discovering semantic places that mainly identify stops along trajectories of movement objects. These clustering algorithms use a parameter set that is tuned in an application-driven way. Beyond the implementation, modification and testing of the algorithms, an additional task of this research work will be the experimentation with parameter tuning. 

Tasks and time schedule

Literature review: on semantic trajectories modelling and discovery of semantic places. 

Pre-processing of data: filter spatial trajectories to reduce measurement noise (outlier removal)

Algorithms’ implementation: implement known algorithms and modify them according to the application needs.

Experiments and testing

Writing thesis documentation

Data and Tools

Dataset: GPS Trajectories Data Set (an open dataset is provided)

Tools: PostgreSQL database with PostGIS extention (tutorial of installation and usage is provided)

Prerequisites

Programming: knowledge of a programming language as the tasks are heavily dependent on implementing algorithms

Data Analysis: prior experience in data analysis is desirable but not mandatory

Interest and willingness to work with movement data and explore moving objects’ behaviour

Contact

M.Sc. Stefania Zourlidou - zourlidou@ikg.uni-hannover.de - Tel. 0511.762-19435 - Office 615
- Institut für Kartographie und Geoinformatik, Appelstraße 9a, 30167 Hannover