I am a Ph.D. candidate at professor dr. Jack van Wijk's visualization group of the faculty of Mathematics and Computer Science at the Eindhoven University of Technology under the supervision of dr.ir. Huub van de Wetering. Our research is part of the Metis project of the Embedded Systems Institute, in cooperation with Thales Nederland BV. The project focuses on finding, analyzing, describing and visualizing uncertainty of heterogeneous information in systems-of-systems in the context of maritime safety and security systems. Our research focuses on the visualization of heterogeneous and uncertain information, specifically maritime data.
Before starting on Metis, I was involved, as master student and later as researcher, in the Poseidon project. Metis is the follow up project to Poseidon.
We present a visualization method for the interactive exploration of predicted positions of moving objects, in particular, ocean-faring vessels. Two simple prediction models, one based on similarity to historical trajectories and one on Monte Carlo simulation, are presented. The prediction models generate temporal probability density fields starting from a known situation. We use contours to visualize spatio-temporal zones of these density fields. Predictions are split into a configurable number of segments for which we render one or more contours. Users, investigating and exploring the possible development of a situation, can see where a vessel will be in the near future according to a given prediction model. Through a number of real-world use cases and a discussion with users, we show our methods can be used in monitoring traffic for collision avoidance, and detecting illegal activities, like piracy or smuggling. By applying our methods to pedestrian movements, we show that our methods can also be applied to a different domain.
Left: The interaction between predictions of two vessels. Both predictions are over 12 minutes, divided into two time intervals. The interaction chance is shown using a discrete color map ranging from blue (low chance), to red (high chance). We see a large area of potential interaction in the second time interval. The range of the color map can be changed to suppress low probabilities. Right: Prediction of the interaction between a tanker and a vessel suspected of piracy. The prediction is divided into three time instances, 6 minutes apart. There are two areas where the pirate vessel may strike; around 12 minutes and around 18 minutes.
In this research we focused on visualizing large scale, multivariate, moving object data, in particular maritime data. These moving objects generally have a large amount of attributes besides time and position, i.e., size, type, velocity, etc. The challenge is how to convey these additional attributes to a user. We researched a method to interactively explore multiple attributes in trajectory data using density maps, i.e., images that show an aggregate overview of massive amounts of data.
Individual trajectories are convolved by moving a smoothing kernel over the trajectories with the speed of the vessel. The smoothed trajectories are then aggregated into a 'density field', which is then visualized using a combination of color mapping and illumination. The user can compute density fields subsets of the data based on any of its attributes using different parameters such as the size of the smoothing kernel and the weight applied to the resulting density. Using a widget, called a distribution map, the user can interactively define subsets in an effective and intuitive way. These density fields can be combined into a single density map where a multi hue color map is applied or a single hue color map is applied to each individual density field and the resulting images are blended. We use four different blending methods:
One of the things we can do with the above method is visual, spatial anomaly detection. We take a large historic set of trajectories as reference and a live data set and aggregate these. A spatial anomaly is then defined as a 'live' vessel moving in an area where there is little or no traffic in the historic set. We use a density field of the historic data only for the illumination to serve as context and show anomalies with a green to red color map as shown in the image above.
As an extension to the multivariate density maps shown above, we have created a visualization scheme in which the user can build a network of parameterizable 'blocks' that can each perform a class of functions. This allows an analyst to encode their domain knowledge into a network of these blocks that can then be used by an operator in an operational context. We have six types of blocks:
An analyst can build networks for a large variety of use cases such as the extraction of busy shipping lanes from a large set of trajectories (see above: a), making clearer images of areas with slow moving vessels by inversely scaling the kernel size by an underlying density map (see above: b) or by finding complex movement patterns such as areas with vessel movements that are moving against the 'main flow' (see above: c).