Data Fusion through Fuzzy-Bayesian Networks for Belief Generation in Cognitive Agents

Abstract

Situation Awareness provides a theory for agents decision making to allow perception and comprehension of his environment. However, the transformation of the sensory stimulus in beliefs to favor the BDI reasoning cycle is still an unexplored subject. Autonomous agent projects often require the use of multiple sensors to capture environmental aspects. The natural variability of the physical world and the imprecision contained in linguistic concepts used by humans mean that sensory data contain different types of uncertainty in their measurements. Thus, to obtain the Situational Awareness for Agents with physical sensors, it is necessary to define a data fusion process to perform uncertainty treatment. This paper presents a model to beliefs generation using fuzzy-bayesian inference. An example in robotics navigation and location is used to illustrate the proposal.