A Monte Carlo simulator for evaluation of AUV configuration in object search and classification missions

A simulation system for evaluation of AUV configurations in search and classification missions has been developed. In particular, the simulator relies on a database that stores information on AUV class, energy storage, navigation system, acoustic sonar payload. The user can combine different AUV configurations by selecting different payloads/energy sources/navigation systems among the ones available in the database and assigning them to an AUV. The acoustic payloads are characterized by a series of data in order to allow evaluation of the detection/classification capability for a given set of mission parameters (vehicle depth, target orientation, bottom typology). A GIS-based interface allows us to define, directly through a map interface, a specific mission scenario, including bathymetry and bottom typology, and it allows us to insert a set of objects within the scenario. The system simulates the mission of the vehicle with the chosen configuration over the defined area, producing as final result the percentage of correctly classified objects and the mission time. The simulation is repeated in a Monte Carlo fashion in order to estimate the expected value of the classification percentages and of the mission time; both estimates are given with a level of confidence measured by use of the Chernoff bound, which links the number of independent runs of the same mission with the probability of the estimate.