Researchers from NSD’s Applied Nuclear Physics Program recently visited the Nevada National Security Site (NNSS) to perform data collections in support of two projects focused on using computer vision technology to automatically identify nuclear material containers. The Container Counting project has developed software that analyzes camera images to automatically locate specific instances of nuclear material containers within a three-dimensional model. The Inventory Verification through Detectors on Robotic Inspection Platforms (IV-DRIP) project has begun leveraging the products from the Container Counting project and coupling them with radiological measurements and robotic autonomy.
The primary goal of the NNSS visit was to acquire data to validate the counting and container-type classification techniques for the Container Counting project, and to collect data for inventory verification with nuclear radiation mapping, both in handheld mode and on robotic platforms (mainly Unmanned Aerial Systems (UAS) and Boston Dynamics’ quadruped robot SPOT dog). The data collection was performed at container storage facilities and the low level nuclear waste facility.
The team utilized machine learning to perform object recognition and enable the robotic systems to autonomously navigate to, and investigate, radiation sources in the nuclear material containers. The two platforms were mounted with ANP-developed sensor systems referred to as Localization And Mapping Platforms (LAMPs) [1], allowing for precise detection of radioactive materials within a given area. The team’s visit to NNSS not only enhanced existing measurement capabilities but also demonstrated the effectiveness of automation in nuclear safeguards scenarios. The team looks forward to further advancements and continued collaboration with NNSS in the future.
The Container Counting and IV-DRIP projects are supported by the Department of Energy National Nuclear Security Administration (NNSA), Defense Nuclear Nonproliferation Research and Development (DNN R&D) Nuclear Safeguards portfolio. The Container Counting and IV-DRIP projects are led by Marco Salathe and Brian Quiter, respectively. Marco Salathe, Brian Quiter, Karthika Balan, and Emil Rofors took part in the measurement campaign at NNSS.
Figure 1: ANP researcher Karthika Balan operating a SPOT robot dog equipped with an ANP-developed multi-sensor system.
Figure 2: A multi-sensor radiation detection system mounted on a Unmanned Aerial System surveys a stack of containers.
References
[1] R. Pavlovsky et al., 3-D Radiation Mapping in Real-time with the Localization and Mapping Platform from Unmanned Aerial Systems and Man-portable Configurations, arXiV:1901.05038 (2018)