Isolating Wildfires using a Convolutional Neural Network based Multi-Agent System

Rocholl, Niels (2021) Isolating Wildfires using a Convolutional Neural Network based Multi-Agent System. Bachelor's Thesis, Artificial Intelligence.

Preview	Text Bachelors_Thesis_Niels_Rocholl_Final.pdf Download (719kB) | Preview
	Text toestemming.pdf Restricted to Registered users only Download (115kB)

Abstract

This study tests the ability of a convolutional neural network (ConvNet) based multiagent system (MAS) to isolate wildfires in a simulation. The ConvNet is designed to map the environmental input to useful moves. The MAS consists of either two or four agents. All systems are tested in grid environments of size 21 × 21, 41 × 41, and 61 × 61. All environments are simulated with a normal and a fast fire propagation speed. Furthermore, a single agent system is implemented for the purpose of comparison. The results show that the four agent system outperforms the single-agent system in all environments. Moreover, the four agent system is able to isolate more than 99% of fires in all environments except one. In the 61 × 61 environment with a fast fire propagation speed, the four agent system is able to isolate 90.5% of the fires on average.

Item Type:	Thesis (Bachelor's Thesis)
Supervisor name:	Wiering, M.A.
Degree programme:	Artificial Intelligence
Thesis type:	Bachelor's Thesis
Language:	English
Date Deposited:	05 Feb 2021 12:42
Last Modified:	05 Feb 2021 12:42
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/23913

Actions (login required)

View Item