UTILIZING THE HRNET ALGORITHM FOR AIMPOINT SELECTION AND POSE ESTIMATION IN HIGH-ENERGY LASER SYSTEMS
Loading...
Authors
Gaines, Gregory
Subjects
High-Resolution Network
HRNet
deep learning
DL
laser weapon systems
LWS
unmanned aerial vehicles
UAV
Naval Postgraduate School
NPS
Spacecraft Research and Design Center
SRDC
percentage of correct keypoints
PCK
Direct Linear Transform
DLT
pose estimation
aimpoint selection
high energy laser
artificial intelligence
machine learning
neural networks
keypoint detection
HRNet
deep learning
DL
laser weapon systems
LWS
unmanned aerial vehicles
UAV
Naval Postgraduate School
NPS
Spacecraft Research and Design Center
SRDC
percentage of correct keypoints
PCK
Direct Linear Transform
DLT
pose estimation
aimpoint selection
high energy laser
artificial intelligence
machine learning
neural networks
keypoint detection
Advisors
Agrawal, Brij N.
Kim, Jae Jun
Date of Issue
2024-06
Date
Publisher
Monterey, CA; Naval Postgraduate School
Language
Abstract
This thesis explores the use of the High-Resolution Net algorithm, a deep learning model for keypoint detection, to enhance automatic target aimpoint selection in laser weapon systems targeting unmanned aerial vehicles. Given the rising use of these systems in modern warfare, advanced technologies for improved targeting accuracy are essential. The study integrates the algorithm to boost identification and targeting efficiency. Utilizing datasets from the Naval Postgraduate School's Spacecraft Research and Design Center, the research evaluates the training and validation of the model. The performance assessment includes metrics like correct keypoint detection percentage, confusion matrices, and precision-recall values. The study addresses challenges such as wing inversion and pose estimation errors in the direct linear transformer and examines filtering techniques to refine the pose estimation accuracy. The results reveal the algorithm performed well overall with keypoint detection, achieving a high accuracy rate for both datasets; however, pose estimation was less accurate, causing wing inversion in the results. Filtering techniques implemented in data post-processing significantly improved these pose estimation results. Future research could explore more diverse algorithms and post-processing filtering techniques. This research marks a significant step towards enhancing targeting accuracy, offering valuable insights for advancements in defense applications.
Type
Thesis
Description
Series/Report No
Department
Organization
Identifiers
NPS Report Number
Sponsors
Funding
Format
Citation
Distribution Statement
Distribution Statement A. Approved for public release: Distribution is unlimited.
Rights
This publication is a work of the U.S. Government as defined in Title 17, United States Code, Section 101. Copyright protection is not available for this work in the United States.