Softec: Optimizing winter road maintenance (industry PhD) 

Our common PhD project with Softec addresses new approaches to predictive modeling and optimization algorithms with a focus on machine learning and artificial intelligence methods. The application domain is winter road maintenance. We work together on reducing situations when the roads are dangerous to drive on and the pace of traffic slows down considerably.

In order to make the winter commutes safer and faster for the drivers, we work on novel ways of improving the current winter road maintenance systems, which are based on data analysis and machine learning techniques. 

Softec collects data from customized IoT devices that monitor the road network. These data together with weather data from SHMÚ will be used in the optimization proposal of dispatching the maintenance vehicles. The environmental impact is also important in this regard.

In this project we focus on three main research areas – providing a more accurate model of the current state of the roads, improving the forecast and simulation of the future state of the roads and developing a way to optimally manage the available resources. 

The first step is to improve precipitation nowcasting based on radar data. To handle this task we use artificial neural networks. This could help the dispatchers to make better proactive decisions and eventually serve as an input to a fully automated management system. 

Prediction methods, which are being developed within this project, are based on the novel trends in neural networks. We will verify them in the given domain, but we believe that our results will be applicable on a broader scale in several other domains.


Machine learning is already making its way into more conservative industries such as meteorology or road maintenance. The results we have achieved assure us that we are on the right track.

Ondrej Svačina, Consultant & Matúš Košík, Consultant


Project team

Peter Pavlík
PhD Student
Viera Rozinajová
Lead and Researcher
Anna Bou Ezzeddine
Marek Lóderer
Martin Výboh
Research Intern