Swarm intelligence is a form of artificial intelligence that is inspired by behavior exhibited by social insects as well as bird flocks and wolf packs. This paper seeks to use swarm intelligence as a means of creating a model to coordinate a group of Unmanned Aerial Vehicles (UAVs) in order to suppress the spread of forest fires. A basic overview of the Particle-Swarm Optimization (PSO) and the Ant- Colony Optimization (ACO) heuristcs are provided as an introduction to the field. Digital pheromones are used to represent the fire in an abstract manner as well as providing a means of communication between the UAVs. A multi-agent simulation is developed in order to test the model. Three experiments are conducted to test various parameters used in the UAV model. It is found that the swarm size and the pheromone threshold both have a strong correlation with the end time of the simulation in addition to the size of the fire. Each experiment is run with random and non-random environmental conditions to create a basis for comparison. In the worst case, the UAV swarm was able to contain the fire 72.3% of the time while the best case contained the fire 90.9% of the time.
Norris, Daniel, "We Didn't Start the Fire! Applying Swarm Intelligence to Unmanned Aerial Vehicles for Forest Fire Suppression" (2010). Senior Independent Study Theses. Paper 808.
Artificial Intelligence and Robotics
Bachelor of Arts
Senior Independent Study Thesis
© Copyright 2010 Daniel Norris