Abstract
Localizing a radiation source in an urban environment is a challenge in nuclear nonproliferation and radiation detection. Mobile radiation sensor networks can be sent into an area of interest to collect count rate measurements at many locations in an attempt to find the radiation source. However, there are numerous factors that cause fluctuations in background count rate such as time and location, and these factors make it challengingto analyze thecollected data. Displaying these measurements in a virtual model has contextualized them and made it easier to determine if environmental factors are introducing false positives. A virtual model has been shown to display measurements in near real-time, but not in real-time. The ability to quickly localize a source is hamperedby this lack of real-time data access as well as by poor qualityofvirtual models. This paper introduces, 1)a new data streaming approach to displaysensor network detector data viewable in a virtual reality model within a second of measurement,2)an application designed to quicken statistics-baseddecision-making on the possibilityof a source in an area, and 3) results of a photogrammetry 3D modeling technique that more easily creates higher fidelity models. Data streaming is improved by using AWS DynamoDB for quick data storage and access. An augmented reality application guides users through the process of sampling an area and statistically determining whether a source is present. Drone photogrammetry via Agisoft Photoscan software is used for high fidelity 3D model creation.
Included in
Radiation Detection, Measurement, and Visualization Assisted by Virtual and Augmented Reality
Purdue University, West Lafayette, Indiana
Localizing a radiation source in an urban environment is a challenge in nuclear nonproliferation and radiation detection. Mobile radiation sensor networks can be sent into an area of interest to collect count rate measurements at many locations in an attempt to find the radiation source. However, there are numerous factors that cause fluctuations in background count rate such as time and location, and these factors make it challengingto analyze thecollected data. Displaying these measurements in a virtual model has contextualized them and made it easier to determine if environmental factors are introducing false positives. A virtual model has been shown to display measurements in near real-time, but not in real-time. The ability to quickly localize a source is hamperedby this lack of real-time data access as well as by poor qualityofvirtual models. This paper introduces, 1)a new data streaming approach to displaysensor network detector data viewable in a virtual reality model within a second of measurement,2)an application designed to quicken statistics-baseddecision-making on the possibilityof a source in an area, and 3) results of a photogrammetry 3D modeling technique that more easily creates higher fidelity models. Data streaming is improved by using AWS DynamoDB for quick data storage and access. An augmented reality application guides users through the process of sampling an area and statistically determining whether a source is present. Drone photogrammetry via Agisoft Photoscan software is used for high fidelity 3D model creation.
https://docs.lib.purdue.edu/can/2019/papers/5