Shawn Whitney

Date of Award

12-2022

Document Type

Thesis

Degree Name

Master of Science in Systems Engineering

Department

Department of Systems Engineering and Management

First Advisor

David Jacques, PhD

Abstract

Recent advances in small Unmmaned Aerial Vehicle (UAV) technology reinvigorates the need for additional research into Wide Area Search (WAS) algorithms for civilian and military applications. But due to the extremely large variability in UAV environments and design, Digital Engineering (DE) is utilized to reduce the time, cost, and energy required to advance this technology. DE also allows rapid design and evaluation of autonomous systems which utilize and support WAS algorithms. Modern WAS algorithms can be broadly classified into decision-based algorithms, statistical algorithms, and Artificial Intelligence (AI)/Machine Learning (ML) algorithms. This research continues on the work by Hatzinger and Gertsman by creating a decision-based algorithm which subdivides the search region into sub-regions known as cells, decides an optimal next cell to search, and distributes the results of the search to other cooperative search assets. Each cooperative search asset would store the following four crucial arrays in order to decide which cell to search: current estimated target density of each cell; the current number of assets in a cell; each cooperative asset’s next cell to search; and the total time any asset has been in a cell. A software-based simulation based environment, Advanced Framework for Simulation, Integration, and Modeling (AFSIM), was utilized to complete the verification process, create the test environment, and the System under Test (SUT). Additionally, the algorithm was tested against threats of various distributions to simulate clustering of targets. Finally, new Measures of Effectiveness (MOEs) are introduced from AI and ML including Precision, Recall, and F-score. The new and the original MOEs from Hatzinger and Gertsman are analyzed using Analysis of Variance (ANOVA) and covariance matrix. The results of this research show the algorithm does not have a significant effect against the original MOEs or the new MOEs which is likely due to a similar spreading of the Networked Collaborative Autonomous Munition (NCAM) as compared to Hatzinger and Gertsman. The results are negatively correlated to a decrease in target distributions standard deviation i.e. target clustering. This second result is more surprising as tighter target distributions could result in less area to search, but the NCAM continue to distribute their locations regardless of clusters identified.

AFIT Designator

AFIT-ENV-MS-22-D-041

DTIC Accession Number

AD1189067

Download

Share

COinS