Geometric Algorithms for the Analysis and Visualization of Heterogeneous Spatio-temporal Data


Probe data from vehicles is one of various types of data that concerns traffic and traffic flow. The corresponding GPS-based trajectories allow various types of analyses, but especially the combination with other data like vehicle-based LiDAR or weather data opens up new possibilities in data correction, pattern detection, and visualization. Both the volume and the heterogeneity of the data poses challenges that can be addressed with an algorithmic approach.

More specifically, we will design visualizations of abstract data like traffic flow within the context of 3D city environments, to discover to what extent the 3D context can be exploited without obscuring the flow data. Such visualizations need modelling and geometric algorithms development, and should work in real-time. Furthermore, we will analyse speed patterns that arise in trajectory data in the context of road intersections, weather, visibility, land cover, and various other data sets. The focus is on modelling, efficient algorithms, provable properties and optimality of the output, and visualization of the patterns. Thirdly, we will develop a model for spatial and spatio-temporal data quality and use it to improve the quality of data sets by outlier detection, completion of missing data, and metadata provision, applied to probe and other traffic-related data. It is specifically the objective to improve data sets by using the heterogeneity of the collection of data sets.

Our foundational research is rooted in computational geometry and provides an algorithmic basis for traffic analysis, traffic visualization, and spatio-temporal data quality handling, with further applications to smart cities.





Prof. dr. B. Speckmann

Verbonden aan

Technische Universiteit Eindhoven, Faculteit Wiskunde en Informatica, Informatica