SenseAir

We are interested in viable data-driven methods for predicting and understanding air pollution. In this context, we are investigating the following research directions: (1) To increase the spatial resolution of air pollution information, a larger number of low-cost monitoring stations can be deployed. How can the sensors be calibrated to provide high quality information? How can we further increase the spatial and temporal resolution of the data received? How can the number of sensor stations be reduced without compromising accuracy? How can we incorporate the different mechanisms of transport and pollution, e.g. by including land use data, traffic information and weather data? (2) There is a lot of interest in forecasting air pollution in the short and long term. How can we predict sudden changes in air pollution? (3) There are different mechanisms to influence air pollution. How can we optimally control air purification systems while considering efficiency and resource consumption? In all these investigations, we use a combination of physical modelling and machine learning.