Wildfire Risk Prediction

Unmanned Aerial Vehicle (UAV) applications have witnessed significant advancements across diverse domains, including surveillance, agriculture, disaster management, and environmental monitoring. UAV sensors can gather critical data, enabling real-time decision-making and enhancing operational capabilities.

However, integrating and managing multiple sensors within a UAV platform pose significant challenges for optimal performance and functionality. UAV platforms rely on different types of sensors. Each sensor serves a specific purpose, from capturing high-resolution imagery to detecting environmental variables and identifying objects of interest. Integrating these diverse sensors into a unified system requires addressing complex technical, logistical, and operational challenges to extract significant information can be a daunting task.

Machine learning algorithms can fuse data from multiple sensors, enabling the extraction of meaningful information. Furthermore, machine learning techniques can learn the correlations and patterns across sensor data by training models on integrated datasets, allowing for more accurate and comprehensive analysis. As a result, we can enhance the overall data quality and facilitate better decision-making, enabling a holistic view of the wildfire situation.

Monitoring the risk of wildfires allows for proactive measures to prevent and mitigate their occurrence. By identifying areas with high fire risk, authorities can implement preventive measures such as controlled burns, fuel management, and enforcing fire safety regulations. Monitoring also enables early detection of potential ignition sources, such as lightning strikes or human activities, allowing immediate response and mitigation efforts. By leveraging advanced monitoring technologies and data-driven approaches, authorities can enhance their ability to respond effectively, mitigate risks, and protect both human lives and natural ecosystems from the devastating impact of wildfires. In this project, we target the study of this monitoring scenario.

We aim to create a data collection system capable of dealing with multiplatform and multi-sensor (multispectral imaging, humidity, temperature, and others) data to automatically build artificial intelligence models to predict wildfire risk in a coverage area.

The data collection, data annotation, and data fusion system will handle different categories of data and platforms, as well as be easily integrated with other data systems related to the problem, such as information from satellites, vegetation maps, or other structured and unstructured data.

This project is under the Air Domain Study (ADS) project, a Brazil-Sweden Cooperation in Aeronautics, whose objective is to better understand the future of the air domain.

These activities will gradually expand knowledge and support stakeholder efforts to position themselves in the future landscape of global aeronautics.