Real time determination of dose radiation through artificial intelligence and virtual reality

In the last years, a virtual environment of Argonauta research reactor, sited in the Instituto de Engenharia Nuclear (Brazil), has been developed. Such environment, called here Argonauta Virtual (AV), is a 3D model of the reactor hall, in which virtual people (avatar) can navigate. In AV, simulations of nuclear sources and doses are possible. In a recent work, a real time monitoring system (RTMS) was developed to provide (by means of Ethernet TCP/IP) the information of area detectors situated in the reactor hall. Extending the scope of AV, this work is intended to provide a continuous determination of gamma radiation dose in the reactor hall, based in several monitored parameters. To accomplish that a module based in artificial neural network (ANN) was developed. The ANN module is able to predict gamma radiation doses using as inputs: the avatar position (from virtual environment), the reactor power (from RTMS) and information of fixed area detectors (from RTMS). The ANN training data has been obtained by measurements of gamma radiation doses in a mesh of points, with previously defined positions, for different power levels. Through the use of ANN it is possible to estimate, in real time, the dose received by a person at any position in Argonauta reactor hall. Such approach allows tasks simulations and training of people inside the AV system, without exposing them to radiation effects. (author)