Dmitry Shadrin, who heads the Remote Sensing Data Processing Group at the AI Center, secured third place in the Artificial Intelligence Technologies category of the BRICS Young Innovators competition. The award ceremony took place at the fourth Young Scientists Congress in Sirius, Russia.

Dmitry was the only winner from Russia and received a $10,000 prize for his project leveraging artificial intelligence to predict and manage the risks of wildfires. The award was presented by Andrey Fursenko, an assistant to the President of the Russian Federation, Deputy Prime Minister of the Russian Federation Dmitry Chernyshenko, Minister of Science and Higher Education of the Russian Federation Valery Falkov, and President of the Russian Academy of Sciences Gennady Krasnikov.

“I am proud that our project has received international recognition at such a prestigious level,” Dmitry Shadrin stated. “This award is a testament to the collective efforts of our team and the comprehensive support provided by Skoltech. I am deeply grateful to my colleagues for their contributions and to the Government of the Russian Federation for prioritizing the development of AI technologies.” 

Under Dmitry’s leadership, the group develops solutions as part of a larger multimodal data processing platform. This platform addresses a wide range of tasks using AI, including environmental monitoring, quantitative surface analysis, and disaster prediction. The wildfire prediction module is one of the platform’s central components, created with support from the Analytical Center for the Government of the Russian Federation.

The wildfire monitoring module utilizes Earth observation data, meteorological inputs, terrain information, and population density figures. For model training, 16 data parameters were analyzed, yielding over 60 distinct features, including vegetation indices, evapotranspiration rates, land cover classifications, and topographic data.

The system employs two methodological approaches:

• Wildfire Probability Prediction The model generates forecasts for up to five days ahead with a spatial resolution of at least 0.25 degrees. It incorporates historical meteorological data and surface condition indicators to deliver precise predictions of wildfire probability.
• Spread Assessment of Detected Fires Once a wildfire is detected, the system predicts its spread, estimating direction and speed based on spatial data and weather forecasts. This functionality helps streamline emergency response and optimize resource allocation.

The system has been adapted to integrate with the operational workflows of the Russian Ministry of Emergency Situations (EMERCOM), enabling nationwide daily forecasts. It is patented and demonstrates significant potential for scaling to other countries, particularly BRICS members, which possess vast territories and forest resources.

“The system we’ve developed enables the efficient allocation of resources to prevent and combat wildfires,” Dmitry Shadrin added. “During wildfire seasons, the model provides real-time predictions, helping to prevent disasters and save lives.”