Radiation Sensor Placement using Reinforcement Learning in Nuclear Security Applications

One of the main concerns in nuclear security is the prevention of terroristic activities. One defense line of such attacks is the use of radiation sensor networks that may identify the transport of nuclear materials that may be potential threats. The problem of deploying a radiation sensor is highly challenging with the sensor placement being at the heart of it. In this paper, we employ the use of reinforcement learning in identifying the best positions of a limited number of gamma ray radiation sensors aiming at maximizing the detection probability of a radioactive source. The research carried in this work integrates the use of radiation physics simulations using the Geant4 of a specific area within the campus of University of Texas at San Antonio. The simulated radioactivity of source that may be of potential is used as an input to the reinforcement learning algorithm to identify the best spots for placing our radiation sensors.