Robotic Neutron Detection for Bayesian Source
Localization, Absence Confirmation,
and Template Matching

In scenarios where no neutron sources or significant changes are declared, the ability to characterize a neutron field and identify possible anomalies could be crucial for verifying compliance to safeguards or arms control agreements. This work demonstrates a unified approach using the N-SpecDir Bot, a Neutron-detecting Spectrally and Directionally sensitive robot, for localization of an anomalous source, absence confirmation, and neutron field template matching. Starting from a simplified particle filter, we introduce several improvements which are applicable to potential applications. We generalize the particle filter to account for unknown background radiation and the possibility of zero sources so that it can simultaneously verify the absence of sources or localize an anomalous source while incorporating information from successive measurements as the N-SpecDir Bot explores its search environment. We also propose a method for adapting the particle filter to template matching with prior measurements by redefining the particles to represent the location and magnitude of deviation from the template. Template acquisition, interpolation, and matching is demonstrated with experiments conducted at Princeton Plasma Physics Laboratory.