GASPADA project update: it’s all about movement

What do gnus on the African savanna have in common with cars making their way through Eindhoven? It may sound odd, but to professor Bettina Speckmann and her co-workers in the GASPADA-project gnus and cars are not that different. It’s all about movement data.

Smarter traffic management

Professor Speckmann has been working with movement data in different settings. In the GASPADA-project (Geometric Algorithms for the analysis and visualization of heterogeneous Spatio-temporal Data), the focus originally was on traffic. By cleverly combining information on the speed and location of cars with data on, for example, streetlights and buildings alongside the road, new insights for traffic management can be developed. Data collection in these kinds of projects is not always easy, Speckmann explains: “Trajectory data can be tricky to gather, especially in urban environments. Data often needs to be cleaned before it can be used in further analysis. To clean the data, we have developed what we call a physically realisable approach. This approach takes into account the physical limitations of cars such as their speed and the fact that most cars will not drive off-road. By making use of such factors, we are able to find the largest subtrajectories that make physical sense. This allows us to quickly clean up large data sets.”

Taking an unexpected turn

Unfortunately, the project took an unexpected turn as some things changed in the project consortium. “One of the main partners, Here Global, set up a serious research department in Eindhoven which created interesting opportunities to collaborate. The first year of our collaboration worked out well. However, we then experienced the downside of working with a multinational company when research budgets were cut and eventually their work in Eindhoven was discontinued completely.”

With the changes in the consortium, the focus of the GASPADA-project also changed from downtown Eindhoven to grasslands in Africa. The approach was similar: combining movement data with data on the surroundings. “Instead of streetlights, we use data on things like ground cover and the weather,” explains Speckmann. “Other than that, we are still building complex algorithms from basic building blocks. An example of a building block is an algorithm that measures the distance between two trajectories. The great thing is that we can use these building blocks when working with trajectories in any given situation, ranging from cars in cities to animals on the savanna.”

Collaborators from Wageningen University work on herd animals, like gnus. By analysing the collective behaviour of herds, it is possible to detect disruptions in their normal patterns that may be caused by poachers. Speckmann: “If you track a rhino and at some point, the rhino stops moving you know that the poachers got to it. The data may be correct, but you would be too late. Our approach is different in that we use other species as so-called sentinels: disturbances by poachers change the normal moving patterns of these species. But to know what normal patterns are, you need to take into account environmental variables like ground cover, weather et cetera. The integration of such variables is exactly what we are good at.”

Looking ahead

The change in direction within the project shows that for professor Speckmann and her team, the algorithmic concepts are the same – it is just about finding good partners to make the algorithms work in a meaningful way. As most passionate scientists, Speckmann has some ideas for future developments. “It would be great to develop our library of algorithms further, especially in the direction of group analysis, like we use for gnus. We can also get more from our approach of physically realisable models. We could for example try making the switch from cars to people. Like cars, people have physical limitations that can be used: most of us cannot fly, for example. Using the physically realisable approach could be of interest in smart city research. I live in the centre of Utrecht, so I see the need for smart traffic- and crowd management every day.”

The movement toolkit developed in this project can be found here: https://movetk.win.tue.nl/