Decentralized Data Fusion Algorithm for Multisensor Radiation Source Search

The Global Technology Connection, Inc. (GTC) and Oak Ridge National Lab (ORNL) team has demonstrated the feasibility of decentralized, scalable data fusion algorithms to accurately localize and characterize radiation source(s) in a complex scene. This has been done by combining the raw output of disparate stationary and moving radiation detectors and processing data into usable information. Nuclear source localization is a demanding problem, and the team has identified three significant challenges regarding this problem: limits imposed by the physics of radiation detection (e.g., detector size, collection efficiency, and measurement time), computational limitations of computer hardware (especially difficulty in implementing real-time algorithms), and lack of information regarding the materials at the scene. The STTR team has addressed all three challenges in the proposed algorithms and demonstrated the feasibility of using those algorithms for real-time implementation. This feasibility study was performed on DTRA-provided data collected at the International Village test site on the Kirkland Air Force Base. During Phase I development, this dataset was used to detect and locate multiple sources of different and the same source type. It has been shown that the developed algorithms outperformed current search methods significantly. To investigate the deployability of the proposed technology, the team created a modular Python package that ingests the discrete detector files (.n42 format) and publishes the estimated source type, locations, intensity, and confidence level to a TAK server for quick and easy viewing on ATAK or WinTAK. The Phase II team consists of GTC, ORNL, Applied Research Associates (ARA), and Arktis. ARA and Arktis will bring valuable real-world expertise to the STTR team and the technology transition process. In Phase II, the algorithms will be refined for three strategic deployment scenarios: Defined search path of one or more moving detectors with optional fixed detectors where detectors could be carried by hand, backpack, truck, or unmanned vehicles. Automated guiding algorithm to rapidly search units to a source of interest with the intent of integrating the algorithm into the unmanned ground vehicle (UGV) and unmanned aerial vehicle (UAV) dynamic path planning tools. Fixed detector networks strategically mounted in airports, ports, or dense public areas. In Phase II, the team will optimize the process from data input to publishing results for each scenario to gain confidence in the respective stakeholders, fuse neutron detections, and the scene interaction (ray tracing) will be modified to consider 3D topology in addition to buildings. GTC has submitted a provisional patent application (application # 63/480,117) for its innovative methodology, with the intention of further pursuing a full patent to safeguard the company's intellectual property, ensure its freedom to operate and prevent any infringement by competitors.