Uncertainty evaluation for a Dezert–Smarandache theory-based localization problem

Abstract

Dezert–Smarandache theory (DSmT) is selected as the fusion method in a decision making system. To compare sensor fusion results, an uncertainty analysis is performed at each level of the decision making system. The supervision is based on the Generalized Aggregated Uncertainty (GAU) measure which is a generalization of Aggregated Uncertainty measure, whereas it is applicable for DSmT results. This measure helps to make decisions on better choice of combinations of sensors or methods of fusion. As an experimental study, localization of an object using cameras’ images is selected. Classic DSmT and hybrid DSmT by using extra knowledge, is applied and then uncertainty evaluation is carried out by the GAU uncertainty measure. The final decision in the presented framework has uncertainty less than each sensor’s measurement. By this method, more accurate and less uncertain results in localizing the object with high conflict sensory data are generated.