Home

Introduction & Background

In this project, I aimed to develop a drone-assisted system for fighting wildfires using the Robot Operating System (ROS). Utilizing multiple UAVs, I focused on detecting and monitoring wildfires to facilitate early intervention and suppression. The project leveraged Voronoi Partitioning Algorithm to optimize the deployment of UAVs, ensuring efficient and unbiased coverage of the affected area.

Multiple UAVs have been extensively used to recognize spot fires and screen out of control fire hazards moving toward a structure, fence, forest, or firefighting crew via remote sensing.

Once the drone detects the fire-affected area, the course of action would be to extinguish the fire by using fire extinguishing balls. The ball is made up of Styrofoam which is environmentally friendly and biodegradable. There are currently two brands that dominate this market, namely: The Elide and the AFO.

Voronoi Tessellation

I used Voronoi Tessellation to divide the area into convex polygons, with each UAV situated at the centroid of its respective polygon. This method ensured that each drone covered an equal area, optimizing the search and detection process. The MATLAB simulation demonstrated the effectiveness of this approach with 'n' number of drones.

Drone Simulation

To simulate the UAVs, I used the Hector Quadrotor in the Gazebo environment. This simulation closely represented real-world flight, allowing for the testing of flight algorithms and control approaches. I equipped the quadrotors with thermal cameras to detect high-temperature areas indicative of fires.

Thermal Environment and Cameras

To recreate a virtual wildfire in the simulated environment, I used the blender to create the environment module including wildfire using different texture modules. This was imported into the gazebo environment.

Results

The simulations showed that using Voronoi Tessellation with multiple UAVs significantly improved the efficiency and accuracy of wildfire detection. The Hector Quadrotor equipped with thermal cameras was able to identify fire-affected areas accurately. The system's ability to operate autonomously and in coordination reduced human error and enhanced the effectiveness of wildfire monitoring and suppression.

Conclusion

This project demonstrated the potential of using autonomous UAVs for efficient wildfire detection and suppression. By implementing the Voronoi Partitioning Algorithm and simulating the system in ROS and Gazebo, I was able to create a robust and scalable solution for wildfire management.