Radiation source localization is a common and critical task across applications such as nuclear facility decommissioning, radioactive disaster response, and security. Traditional count-based sensors (e.g. Geiger counters) infer range to the source based on the observed number of gamma photons, expected source strength, and assumed intermediate attenuation from the environment. In cluttered 3D settings, such sensors struggle to efficiently disambiguate between symmetries caused by sensor, source, and environment configurations. The recent commercialization of Compton gamma cameras that can not only record counts but also image the direction of incident gamma photons presents new opportunities for efficient radiation mapping and source localization. By leveraging Compton scattering physics, these sensors can obtain bearing measurements to sources independent of strength or attenuation for efficient and accurate source localization.

This three-part thesis develops a full exploration and source localization framework that uses the observations of a gamma camera to efficiently discover and localize sources in cluttered 3D environments. First, an approach to mapping the spatial distribution of radiation in an environment using a gamma camera is presented where observations along the camera’s trajectory are incorporated into a consistent voxel grid-based map of source occupancy probabilities. Second, the assumption of a predetermined trajectory is relaxed and an active source localization framework is developed that greedily selects new waypoints that maximize the Fisher Information provided by the camera’s range and bearing observations. As the time required for imaging scales inversely with the square of the distance between the source and camera, proposed active source localization framework is appropriate only within reasonable proximity to a source. Thus, a complementary frontier-based exploration method is finally developed that biases the frontier selection by the observed radiation field gradient to quickly search an environment until a proximal source is detected.

Committee:

William “Red” Whittaker (Co-chair)
Nathan Michael (Co-chair)
Michael Kaess
Sankalp Arora