CREDO: Identifizierung von unbekannten Objekten mittels kredaler Klassifizierung
A project funded by the Center for Military and Civilian Systems of the Department of Defence, Civil Protection and Sports of the Swiss Confederation. Modern military identification problems require the fusion of information coming from several sources. In project CREDO, we consider as example the identification of a civil aircraft flying in a no-fly zone established for the protection of a potential target. In this application, information about the aircraft is collected by several sources like for example 2D and 3D radars, TV cameras, observation units and interceptors. The fusion of the information coming from these sources is a difficult task, because their reliability depends on many observable and unobservable factors and because their relative information can be contrasting or incomplete. We model this complex identification problem using a credal network. Credal networks are a recent generalization of Bayesian networks to imprecise probabilities and are suitable for handling incomplete, contrasting and unfeasible information. The resulting model could be useful for training the persons that in practice have to take decisions and to evaluate the efficiency of a given identification architecture. The project CREDO is a preliminary study for evaluating the suitability of a wider application of credal networks to military identification problems.
People Alessandro Antonucci
Alberto Piatti
Marco Zaffalon



Copyright © 2003 - 2012 IDSIA. All Rights Reserved.