Mathematical models to build multi-radiation detector algorithms

RT&L FOCUS AREA(S): Artificial Intelligence/ Machine Learning TECHNOLOGY AREA(S): Nuclear; Sensors OBJECTIVE: Develop flexible radiation algorithms deployed across battlefield networks to enable the linking of multiple detector variants and fusing of raw detector outputs into usable information. DESCRIPTION: Often, multiple detectors, and multiple detector variants are deployed to characterize a complex scene (i.e. stationary detectors, handheld radioisotope devices, vehicle-mounted detectors, and backpack detectors) within 1 square kilometer. This topic seeks to develop flexible radiation detection algorithms leveraging proven mathematical data models that would sit either at a node for multiple detectors or at a command center that fuses raw detector outputs into useable information. Multiple data types are included in this deployment modality: gross gamma/neutron counts, gamma spectral data, GPS data, etc. Advances in big data theory, machine learning, and artificial intelligence have yielded new mathematical models that could be applied to multiple radiation detection sensors to fuse data in a way that novel algorithms may analyze the overall data input, instead of discrete sensor data. The intent of this topics is to leverage these new mathematical principals and models to decrease time to localize and characterize radiological signature anomalies in a complex scene by leveraging data from all radiation detector types. This would serve to better protect warfighters by reducing mission times and provide commanders better mission radiological characterization for the overall scene. PHASE I: Identification of multi-radiation detector algorithms and demonstrate their potential to improve the identification, characterization, and/or localization of a radioactive source in a complex scene as compared to the singular detector algorithm. Multiple candidate algorithms shall be down selected for further development in Phase II. Demonstrate pathways for meeting the Phase II performance goals through feasibility studies at the end of Phase I. PHASE II: Demonstrate enhanced identification, characterization and/or localization of radioactive sources with the multi-detector algorithm that fuses data (gamma and neutron radiation outputs, and GPS location/time) from disparate ground based and mobile detector types. Demonstrate improved performance of the multi-detector algorithm over single-system algorithms. The algorithm should support the integration of additional new detector types. PHASE III DUAL USE APPLICATIONS: Field demonstration in radiation environment with users deploying multiple and varied radiation detectors linked via communications to a network node in which the algorithm receives detector outputs. The algorithm must conduct scene characterization in real-time as operators move through a complex environment with disparate detector modalities. The multi-system algorithm will be directly compared to legacy single-system algorithms to assess impact on mission. Develop commercialization and transition plan to DoD end users. REFENCES: 1. Information Fusion Volume 57, May 2020, Pages 115-129; https://doi.org/10.1016/j.inffus.2019.12.001 ; 2. Joint Pub 3-11