Economic and lightweight radiation mapping: using mmWave radar and lidar-trained Pix2Pix cGAN to create 2D indoor maps for radiation localization

Hannah S. Parrilla; Mark S. Bandstra; Reynold J. Cooper; Yue Shi Lai; Thomas D. MacDonald; Ivo Marković; Victor Negut; Stijn Wielandt

doi:10.1117/12.3025758

7 October 2024 Economic and lightweight radiation mapping: using mmWave radar and lidar-trained Pix2Pix cGAN to create 2D indoor maps for radiation localization

Hannah S. Parrilla, Mark S. Bandstra, Reynold J. Cooper, Yue Shi Lai, Thomas D. MacDonald, Ivo Marković, Victor Negut, Stijn Wielandt

Author Affiliations +

Proceedings Volume 13149, Unconventional Imaging, Sensing, and Adaptive Optics 2024; 1314916 (2024) https://doi.org/10.1117/12.3025758
Event: Optical Engineering + Applications, 2024, San Diego, California, United States

Abstract

Nuclear Scene Data Fusion (SDF) integrates radiation data and other contextual sensor measurements to enable free-moving localization and mapping of nuclear radiation from person or robot-borne sensor systems referred to as Localization and Mapping Platforms (LAMPs). The LAMP sensor suite typically utilizes lidar to create high-fidelity 3D point clouds representing the measurement scene. The high precision and accuracy of lidar has been essential to SDF, but its larger size, weight, power, and cost (SWaP-C) requirements limit its use in lightweight, portable applications, such as drones or handheld systems for remote field operations. In these cases, automotive millimeter-Wave radar presents a cost-effective, lightweight, and energy-efficient alternative, albeit with a significant compromise in spatial resolution and accuracy. We consider utilizing 2D radar point clouds to create occupancy maps of the scene. One method explored generates radar occupancy grids refined by a lidar-trained Pix2Pix conditional Generative Adversarial Network (cGAN) to approach lidar occupancy grid accuracy. We utilize these radar point occupancy grids to locate and quantify a gamma-ray point source in an environment never-before-seen by our lidar-trained Pix2Pix model, and we compare the results to those generated using standard lidar occupancy grids to assess the feasibility of using radar in place of lidar in some situations and environments.

Conference Presentation

(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Hannah S. Parrilla, Mark S. Bandstra, Reynold J. Cooper, Yue Shi Lai, Thomas D. MacDonald, Ivo Marković, Victor Negut, and Stijn Wielandt "Economic and lightweight radiation mapping: using mmWave radar and lidar-trained Pix2Pix cGAN to create 2D indoor maps for radiation localization", Proc. SPIE 13149, Unconventional Imaging, Sensing, and Adaptive Optics 2024, 1314916 (7 October 2024); https://doi.org/10.1117/12.3025758

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available