Evidential Reasoning For Geographic Evaluation For Helicopter, Route Planning

Abstract

In order to plan operations where knowledge of significant elements is imprecise and uncertain, a means of characterizing the situation in terms of the various factors that may influence those operations must be provided. In this paper we discuss an approach to that characterization that uses evidential reasoning to handle the uncertainty, imprecision, and incompleteness typical of sources of real-world information and knowledge, to support planning routes for military helicopters. Evidential reasoning is a maturing collection of inference techniques for reasoning with uncertain information. Based on the Shafer-Dempster theory of evidence, evidential reasoning uses a non-Bayesian updating scheme to combine evidence provided by multiple, diverse knowledge sources. Knowledge sources in an evidential reasoning system are not required to attribute their belief to a universe of discourse comprised solely of mutually exclusive, exhaustive, singleton events, as required by a classical probability approach. Rather, they may express levels of ignorance explicitly by allocating belief to disjunctions of propositions, thereby leading directly to an interval measure of belief; ignorance is expressed by the width of this interval. Evidential reasoning evolved from consideration of appropriate models for reasoning about information acquired from sensors, and therefore seems natural for drawing conclusions from sensor data and prestored maps regarding the degree to which a selected geographic area will support certain activities. Here, we discuss evidential reasoning and illustrate the utility of the technology for classifying geographic areas by describing our current map-and-sensor-based research in which we estimate the utility of land areas for concealing helicopter operations.